CN107529709A - Decoder, encoder, the method and storage medium of decoding and encoded video - Google Patents
Decoder, encoder, the method and storage medium of decoding and encoded video Download PDFInfo
- Publication number
- CN107529709A CN107529709A CN201710694425.0A CN201710694425A CN107529709A CN 107529709 A CN107529709 A CN 107529709A CN 201710694425 A CN201710694425 A CN 201710694425A CN 107529709 A CN107529709 A CN 107529709A
- Authority
- CN
- China
- Prior art keywords
- motion vector
- binary
- vertical component
- horizontal component
- binary digit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 61
- 238000003860 storage Methods 0.000 title claims abstract description 16
- 230000033001 locomotion Effects 0.000 claims abstract description 217
- 239000013598 vector Substances 0.000 claims abstract description 201
- 230000000903 blocking effect Effects 0.000 claims abstract description 34
- 230000036961 partial effect Effects 0.000 claims description 40
- 229910002056 binary alloy Inorganic materials 0.000 claims description 39
- 230000003044 adaptive effect Effects 0.000 claims description 22
- 238000013139 quantization Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 9
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 6
- 239000004744 fabric Substances 0.000 claims description 5
- 239000000872 buffer Substances 0.000 description 110
- 238000009826 distribution Methods 0.000 description 40
- 238000006243 chemical reaction Methods 0.000 description 38
- 238000013507 mapping Methods 0.000 description 36
- 230000008859 change Effects 0.000 description 28
- 230000009466 transformation Effects 0.000 description 28
- 239000012634 fragment Substances 0.000 description 23
- 230000014509 gene expression Effects 0.000 description 18
- 230000004913 activation Effects 0.000 description 14
- 238000004422 calculation algorithm Methods 0.000 description 14
- 230000008569 process Effects 0.000 description 14
- 230000003595 spectral effect Effects 0.000 description 13
- 230000006978 adaptation Effects 0.000 description 12
- 230000005540 biological transmission Effects 0.000 description 12
- 238000012545 processing Methods 0.000 description 11
- 230000002829 reductive effect Effects 0.000 description 10
- 238000013461 design Methods 0.000 description 9
- 235000019580 granularity Nutrition 0.000 description 9
- 230000015654 memory Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 230000008901 benefit Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 230000005764 inhibitory process Effects 0.000 description 7
- 238000005259 measurement Methods 0.000 description 7
- 230000003139 buffering effect Effects 0.000 description 6
- 230000006835 compression Effects 0.000 description 6
- 238000007906 compression Methods 0.000 description 6
- 230000009467 reduction Effects 0.000 description 6
- 230000004048 modification Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 238000011176 pooling Methods 0.000 description 5
- VEMKTZHHVJILDY-UHFFFAOYSA-N resmethrin Chemical compound CC1(C)C(C=C(C)C)C1C(=O)OCC1=COC(CC=2C=CC=CC=2)=C1 VEMKTZHHVJILDY-UHFFFAOYSA-N 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 230000033228 biological regulation Effects 0.000 description 4
- 238000010276 construction Methods 0.000 description 4
- 230000010181 polygamy Effects 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 241000209094 Oryza Species 0.000 description 3
- 235000007164 Oryza sativa Nutrition 0.000 description 3
- 230000003213 activating effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 230000000670 limiting effect Effects 0.000 description 3
- 235000009566 rice Nutrition 0.000 description 3
- 230000001568 sexual effect Effects 0.000 description 3
- 230000003068 static effect Effects 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000009795 derivation Methods 0.000 description 2
- 238000006073 displacement reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000003780 insertion Methods 0.000 description 2
- 230000037431 insertion Effects 0.000 description 2
- 230000008707 rearrangement Effects 0.000 description 2
- 235000015170 shellfish Nutrition 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 230000008093 supporting effect Effects 0.000 description 2
- 230000009885 systemic effect Effects 0.000 description 2
- 235000019587 texture Nutrition 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 241000023320 Luma <angiosperm> Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 210000003484 anatomy Anatomy 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000010009 beating Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000001976 improved effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- OSWPMRLSEDHDFF-UHFFFAOYSA-N methyl salicylate Chemical compound COC(=O)C1=CC=CC=C1O OSWPMRLSEDHDFF-UHFFFAOYSA-N 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000009738 saturating Methods 0.000 description 1
- 230000008054 signal transmission Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- GOLXNESZZPUPJE-UHFFFAOYSA-N spiromesifen Chemical compound CC1=CC(C)=CC(C)=C1C(C(O1)=O)=C(OC(=O)CC(C)(C)C)C11CCCC1 GOLXNESZZPUPJE-UHFFFAOYSA-N 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
- H03M7/40—Conversion to or from variable length codes, e.g. Shannon-Fano code, Huffman code, Morse code
- H03M7/4031—Fixed length to variable length coding
- H03M7/4037—Prefix coding
- H03M7/4043—Adaptive prefix coding
- H03M7/4068—Parameterized codes
- H03M7/4075—Golomb codes
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
- H03M7/40—Conversion to or from variable length codes, e.g. Shannon-Fano code, Huffman code, Morse code
- H03M7/4006—Conversion to or from arithmetic code
- H03M7/4012—Binary arithmetic codes
- H03M7/4018—Context adapative binary arithmetic codes [CABAC]
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
- H03M7/40—Conversion to or from variable length codes, e.g. Shannon-Fano code, Huffman code, Morse code
- H03M7/42—Conversion to or from variable length codes, e.g. Shannon-Fano code, Huffman code, Morse code using table look-up for the coding or decoding process, e.g. using read-only memory
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
- H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/174—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a slice, e.g. a line of blocks or a group of blocks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/184—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being bits, e.g. of the compressed video stream
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/513—Processing of motion vectors
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/513—Processing of motion vectors
- H04N19/517—Processing of motion vectors by encoding
- H04N19/52—Processing of motion vectors by encoding by predictive encoding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/70—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Theoretical Computer Science (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Diaphragms And Bellows (AREA)
- Sealing Devices (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
Abstract
The present invention one decoder of description, encoder, decoding and the method and storage medium of encoded video, the decoder is used for from data stream video, the horizontal component of motion vector difference and vertical component are encoded in the data flow by it using the binarization of horizontal component and vertical component, in the horizontal component less than cutoff and vertical component, and respectively in the first interval in the domain of the prefix for blocking unitary code form in cutoff and the combination in horizontal component and the suffix of the Exp- Golomb form of vertical component, it is being more than and is being equal in the horizontal component of cutoff and the second interval in the domain of vertical component, what the binarization was respectively equal to horizontal component and vertical component blocks unitary code, wherein cutoff is 2 and Exp- Golomb has order 1.It includes：Entropy decoder, inverse symbolism device and reconstructor.
Description
Present patent application is to propose within 06 18th, 2012 international application, enters Chinese state within 02 14th, 2014
Family's stage, entitled " entropy code of motion vector difference ", the division of the patent of invention of Application No. 2012800398963
Application.
Technical field
The present invention relates to a kind of entropy code concept of encoded video data.
Background technology
Various video codec as is generally known in the art.Generally, these codecs reduce required data volume
To represent video content, i.e. these codecs are compressed to data.It is well-known in the context of Video coding
The compression of video data can be advantageously carried out by order using different coding techniques：Using motion compensated prediction so as to
Predicted pictures content.The motion vector and prediction residual determined in motion compensated prediction need to carry out lossless entropy code.In order to
Data volume is further reduced, motion vector needs to be predicted in itself so that only represent the motion of motion-vector prediction residual error
Phasor difference needs to carry out entropy code.In h .264, for example, using the program summarized just now to transmit on motion vector difference
Information.Specifically, motion vector difference is binarized into blocking unitary code and index brother's human relations from some cutoff
Bit stream corresponding to the combination of cloth code.When the binary file of Exp- Golomb is carried out simply using equiprobability bypass mode
Coding, and fixation probability be 0.5 when, provide several contexts for the first binary file.Cutoff is chosen for nine.Accordingly
Ground, there is provided a large amount of contexts are poor with encoding motion vector.
But, there is provided a large amount of contexts not only increase codec complexity, but also negative shadow is generated to code efficiency
Ring：If few access context, the adaptation of probability, i.e. during entropy code with each context-sensitive probability
The adaptation of estimation can not be carried out effectively.Correspondingly, the probability Estimation of misapplication is estimated actual symbol statistics.Moreover,
Several contexts are provided if some binary file of binarization, then the selection wherein carried out may need to check it
Necessity may hinder adjacent binary file/syntax element value of perform decoding technique.On the other hand, if set
Very little, then the binary file of the actual symbol statistics of height change is grouped amount of context together in a context, and
And correspondingly, can not efficient coding binary file associated there with this context-sensitive probability Estimation.
Need the code efficiency of the entropy code of lasting further raising motion vector difference.
The content of the invention
It is therefore an object of the present invention to provide such Coded concepts.
The purpose is realized by the theme for the independent claims enclosed with it.
According to the one side of the application, there is provided a kind of decoder being used for from data stream video, utilize level point
The horizontal component of motion vector difference and vertical component are encoded in data flow by amount and the binarization of vertical component, less than
The horizontal component of cutoff and vertical component, and respectively in the prefix for blocking unitary code form of cutoff and in level
In the first interval in the domain of the combination of the suffix of the Exp- Golomb form of component and vertical component, cut being more than and being equal to
In the horizontal component and the second interval in the domain of vertical component being only worth, binarization is respectively equal to horizontal component and vertical component
Block unitary code, wherein cutoff is 2 and Exp- Golomb has order 1, including：Entropy decoder, it is configured to, for
The horizontal component and vertical component of motion vector difference, drawn and blocked from data flow using context adaptive binary entropy decoding
Unitary code, and each binary digit position for blocking unitary code has what a proper context, its level to motion vector difference
It is common for component and vertical component, and Exp- Golomb utilizes constant equiprobability bypass mode to be moved
The binarization of phasor difference；Inverse symbolism device, it is configured to the binarization of the syntactic element of inverse binarization motion vector difference
To obtain the integer value of the horizontal component of motion vector difference and vertical component；Reconstructor, it is configured to the water of motion vector difference
The integer value of the amount of dividing equally and vertical component rebuilds video.
According to further aspect of the application, there is provided a kind of encoder for being used to encode video into data flow, including：
Constructor, it is configured to by motion compensated prediction come predictably encoded video, motion compensated prediction is using motion vector and passes through
Predicted motion vector sets the horizontal component of motion vector difference and the integer value of vertical component to carry out predictably encoding motion vector
To represent the prediction error of predicted motion vector；Symbolism device, binarization integer value is configured to obtain motion vector difference
The binarization of horizontal component and vertical component, in the horizontal component less than cutoff and vertical component, and it is in respectively
The prefix for blocking unitary code form of cutoff and in the suffix of horizontal component and the Exp- Golomb form of vertical component
In the first interval in the domain of combination, it is being more than and is being equal in the horizontal component of cutoff and the second interval in the domain of vertical component,
What binarization was respectively equal to horizontal component and vertical component blocks unitary code, and wherein cutoff is 2 and Exp- Golomb has
There is order 1；And entropy coder, be configured to, for the horizontal component and vertical component of motion vector difference, using context from
Adaptation binary system entropy code will block unitary code and be encoded into data flow, and each binary digit position for blocking unitary code has
What a proper context, it is common, and Exp- Golomb for horizontal component and vertical component to motion vector difference
Coding utilizes constant equiprobability bypass mode.
According to having on one side for the application, there is provided a kind of method being used for from data stream video, utilize level point
Amount and the binarization of vertical component encode the horizontal component of motion vector difference and vertical component to data flow, less than cutting
The horizontal component that is only worth and vertical component, and respectively in the prefix for blocking unitary code form of cutoff and in level point
In the first interval in the domain of the combination of the suffix of the Exp- Golomb form of amount and vertical component, it is being more than and is being equal to cut-off
In the horizontal component of value and the second interval in the domain of vertical component, binarization is respectively equal to horizontal component and vertical component
Unitary code is blocked, wherein cutoff is 2 and Exp- Golomb has order 1, including：For the level point of motion vector difference
Amount and vertical component, are drawn from data flow using context adaptive binary entropy decoding and block unitary code, and block unitary code
Each binary digit position there is what a proper context, for its horizontal component and vertical component to motion vector difference
It is common, and Exp- Golomb utilizes constant equiprobability bypass mode to obtain the binarization of motion vector difference；It is inverse
The binarization of the syntactic element of binarization motion vector difference is to obtain the horizontal component of motion vector difference and vertical component
Integer value；The integer value of horizontal component and vertical component based on motion vector difference rebuilds video.
According to the another aspect of the application, there is provided a kind of encoder for being used to encode video into data flow, including：
By motion compensated prediction come predictably encoded video, motion compensated prediction utilizes motion vector and passes through predicted motion vector
Transported with the horizontal component of setting motion vector difference and the integer value of vertical component come predictably encoding motion vector with representing to predict
The prediction error of dynamic vector；Binarization integer value is to obtain the binary system of the horizontal component of motion vector difference and vertical component
Change, in the horizontal component less than cutoff and vertical component, and be in the prefix for blocking unitary code form of cutoff respectively
In the first interval in the domain of the combination in horizontal component and the suffix of the Exp- Golomb form of vertical component, more than
And equal to cutoff horizontal component and vertical component domain second interval in, binarization be respectively equal to horizontal component and
Vertical component blocks unitary code, and wherein cutoff is 2 and Exp- Golomb has order 1；And for motion vector difference
Horizontal component and vertical component, will block unitary code using context adaptive binary entropy code and be encoded into data flow, and
The each binary digit position for blocking unitary code has what a proper context, its horizontal component to motion vector difference and hangs down
It is common for straight component, and Exp- Golomb coding utilizes constant equiprobability bypass mode.
The present invention's is found to be substantially：The code efficiency of the entropy code of motion vector difference can be further by with lower section
Formula will be used to block unitary code and be down to two so as to the cutoff of binarization motion vector difference and blocked so that only existing to improve
Two binary digit positions of unitary code, if order one is used for Exp- Golomb so that binarization is from cutoff
Motion vector difference, and if in addition, be respectively that two binary digit positions for blocking unitary code provide a proper context,
Then so that the context selection of binary digit or syntax element value based on adjacent image block is It is not necessary to and avoid these
The binary digit of binary digit position is subdivided into very much context, so as to probability adaptation proper function, and it is if identical
Context be used for horizontal component and vertical component, then thus further reduce the negative effect of too fine context subdivision.
Further, as a result find, sweared when that will set with predicted motion vector and the desired amount of motion to be transmitted of reduction
When the advanced method of amount difference is combined together, the setting of the entropy code on motion vector difference referred to just now is particularly useful.Example
Such as, multiple motion vector predictors can be provided to obtain the sorted lists of motion vector predictor, and fortune can be used
The index of dynamic vector prediction value list is to determine actual motion vector predictor, the prediction of the actual motion vector predictor
Residual error is represented with the motion vector difference in discussing.Although the information on used list index must be from the number of decoding side
Drawn according to stream, but to improve the macro-forecast quality of motion vector, and correspondingly, the size of motion vector difference is further subtracted
It is small so that in general further to improve code efficiency, and reduce the horizontal component of cutoff and motion vector difference and vertical
The motion-vector prediction for being used in conjunction with being suitable for this improvement of the context of component.On the other hand, can use merge so as to
Reduce the quantity for the motion vector difference to be transmitted in data flow.It is grouped into therefore, pooling information can be transferred in signal
Transmitted in the data flow of the decoder block of the block subdivision of one group of block.Motion vector difference then can be with these merging groups
It is that unit is transmitted in data flow for unit rather than independent blocks, thus reduces the quantity for the motion vector difference that must be transmitted.
Because the cluster of block reduces the interior correlation between adjacent motion vectors difference, the omission referred to just now is one two
System position position provides several contexts and prevents entropy code scheme too thin to context progress according to adjacent motion vectors difference
Classification.On the contrary, merge concept using the interior correlation between the motion vector difference of adjacent block, and correspondingly, one
One context of binary digit position, i.e. the context of horizontal component and vertical component is enough.
Brief description of the drawings
Hereinafter, the preferred embodiment of the application is described with reference to the accompanying drawings, in the accompanying drawings
Fig. 1 shows the block diagram of the encoder according to embodiment；
Fig. 2A -2C are schematically shown is subdivided into different blocks by array of samples such as picture；
Fig. 3 shows the block diagram of the decoder according to embodiment；
Fig. 4 illustrates in greater detail the block diagram of the encoder according to embodiment；
Fig. 5 illustrates in greater detail the block diagram of the decoder according to embodiment；
Fig. 6 schematically shows conversion of the block from spatial domain to spectral domain, resulting transform blockiis and its becomes again
Change；
Fig. 7 shows the block diagram of the encoder according to embodiment；
Fig. 8 is shown is suitable for what bit stream caused by the encoder as Fig. 8 was decoded according to embodiment
The block diagram of decoder；
Fig. 9 is the schematic diagram for showing the packet with polynary partial bit stream according to embodiment；
Figure 10 be show according to further embodiment have using fixed dimension fragment can selected episode packet
Schematic diagram；
Figure 11 shows the decoder of the support pattern switching according to embodiment；
Figure 12 shows the decoder of the support pattern switching according to further embodiment；
Figure 13 shows the encoder of the decoder of suitable Figure 11 according to embodiment；
Figure 14 shows the encoder of the decoder of suitable Figure 12 according to embodiment；
Figure 15 shows pStateCtx and fullCtxState/256**E** mapping；
Figure 16 shows decoder according to an embodiment of the invention；
Figure 17 shows encoder according to an embodiment of the invention；
Figure 18 schematically shows motion vector difference binarization according to embodiments of the present invention；
Figure 19 schematically shows the merging concept according to embodiment；
Figure 20 schematically shows the motion-vector prediction scheme according to embodiment.
Embodiment
It should be noted that during the description of accompanying drawing, the element in several width figures of these figures is appeared in these figures
Represented with the same references in each, as the feature of these elements, repetitive description will be avoided so that avoid need not
The repetition wanted.However, unless specifically stated, the feature and description otherwise provided relative to a width figure will
Suitable for other accompanying drawings.
Hereinafter, the embodiment of overall Video coding concept is described first against Fig. 1 to Figure 10.Fig. 1 to Fig. 6 is related to
And the part of the coding and decoding video operation on syntactic level.Following Fig. 8 to Figure 10 is related to is changed to data with syntactic element circulation
Stream and stream compression are changed to the embodiment of the part of the relevant code of syntactic element stream.Therefore, with for Fig. 1 to Figure 10 typical cases
The specific aspect and embodiment of the form in the cards description present invention of the universal of ground general introduction.
Fig. 1 shows the example of encoder 10, wherein the various aspects of the application can be realized.
Message sample array 20 is encoded to data flow by encoder.Message sample array can represent for example with lightness
(brightness) message sample corresponding to value, color (color) value, brightness (luma) value, colourity (chroma) value etc..So
And in the case where array of samples 20 is the depth map of the generation such as by optical sensor, message sample can also be deep
Spend (depth) value.
Encoder 10 is the encoder based on block.That is, encoder 10 in units of block 40 by array of samples
20 are encoded to data flow 30.Coding is carried out in units of block 40, and to be not necessarily referring to encoder 10 right completely independently of each other
These blocks 40 are encoded.On the contrary, encoder 10 can use the reconstruction of the block of previous coding so as to extrapolation
(extrapolate) can be compiled or infra-frame prediction residue block, and using the granularity (granularity) of block to set
Code parameter, that is, set the mode that pair array of samples region corresponding with each block is encoded.
Further, encoder 10 is transform coder.That is, encoder 10 brings code area by using change
Block 40 from spatial domain by the message sample in each block 40 to be delivered to spectral domain.It can use two-dimensional transform such as FFT's
DCT etc..Preferably, block 40 has secondary square shape or rectangular shape.
Array of samples 20 be subdivided into block 40 be for illustration purposes only shown in Fig. 1.Fig. 1 shows array of samples
20 be subdivided into each other with non-overlap mode against quadratic power or rectangle block 40 conventional two-dimensional arrangements.The chi of block 40
It is very little to predefine.That is, encoder 10 cannot be by the letter on the resource block size of the block 40 in data flow 30
Breath is transferred to decoding side.For example, decoder is expected predetermined resource block size.
However, it is possible to some alternatives be present.For example, block can overlap each other.However, the overlap operation may
Such a degree is confined to, i.e., each block has not by the overlapping part of any adjacent block, or causes the every of block
Individual Sample Maximal limit is by an overlapping block in the adjacent block that is arranged side by side along predetermined direction and current block.The latter
Refer to that the adjacent block of the right and left can be with overlapping current block all to cover current block, but they not can that
This is overlapping, and such case is equally applicable to the neighbouring part on Vertical Diagonal line direction.
As further alternative, from array of samples 20 to the subdivision of block 40 can by encoder 10 using via
Bit stream 30 is transferred to the subdivided information of the used subdivision of decoder-side and the content of array of samples 20 is adapted.
Fig. 2A -2C show the different instances that array of samples 20 is subdivided into block 40.Fig. 2A is shown based on quaternary tree
Array of samples 20 is subdivided into various sizes of block 40, representational block 40a, 40b, the 40c with increased size
And 40d instructions.According to Fig. 2A subdivision, array of samples 20 is first divided into the tree block 40d of conventional two-dimensional arrangements, the tree
Thus block 40 has independent subdivided information related to this, some tree block 40d accordingly can be according to quad-tree structure
Carry out further subdivision or without subdivision.Tree block to the block 40d left sides is exemplarily subdivided according to quad-tree structure
For smaller block.Encoder 10 can become to carrying out a two dimension with solid line and each block shown in phantom in fig. 2
Change.In other words, encoder 10 by block segment in units of carry out transform array 20.
Can be using more generally replacing the subdivision based on quaternary tree based on the subdivision set more, and each level
The quantity of child node can be different between different levels.
Fig. 2 B show another example of subdivision.According to Fig. 2 B, array of samples 20 is first split into mutually supporting with non-overlapped
The mode leaned on is arranged to the macro zone block 40b of conventional two-dimensional arrangements, wherein each macro zone block 40b makes subdivided information related to this,
Macro zone block is not subdivided accordingly, or if subdivision, then with conventional two-dimensional approach be subdivided into equal sized sub-block so as to up to
To the different subdivision granularities of different macro zone blocks.Result is to segment the array of samples 20 in various sizes of block 40, different sizes
Expression 40a, 40b and 40a ' instruction.As with Fig. 2A, encoder 10 in Fig. 2 B to using solid line and shown in phantom
Each block carry out two-dimensional transform.Fig. 2 C will be discussed later.
Fig. 3, which is shown, can decode that the data flow 30 generated by encoder 10 with the reconstructed version of reconstruction sample array 20
60 decoder 50.Decoder 50 extracts the conversion coefficient block of each block 40 from data flow 30 and by each change
Change coefficient blocks and carry out inverse transformation to rebuild rebuild version 60.
Encoder 10 and decoder 50 are configurable to carry out entropy coding/decoding so as to respectively will be on conversion coefficient area
The information insertion data flow of block simultaneously extracts the information from data flow.This relevant side according to different embodiments will be described later
The details in face.It should be noted that data flow 30 not necessarily includes the conversion coefficient area of all blocks 40 on array of samples 20
The information of block.On the contrary, the subset of block 40 can be encoded to bit stream 30 in another way.For example, encoder 10 may decide that
Avoid that bit stream 30 will be inserted for the conversion coefficient block of some block of block 40, but decoder 50 will be enable pre-
Survey or the optional coding parameter of each block in filling reconstructed version 60 inserts bit stream 30 to replace in addition.For example, encoder
10 can carry out texture analysis so as to which block is positioned in array of samples 20, and the array of samples 20 can lead in decoder-side
Decoder is crossed by textures synthesis to fill and accordingly be indicated in the bitstream.
As discussed for figure below, conversion coefficient block not necessarily represents the original of each block 40 of array of samples 20
The spectral domain of message sample represents.On the contrary, this conversion coefficient block can represent that the spectral domain of the prediction residual of each block 40 represents.
Fig. 4 shows the embodiment of this encoder.Fig. 4 encoder includes conversion stages 100, entropy coder 102, inverse transformation stage
104th, fallout predictor 106 and subtracter 108 and adder 110.Subtracter 108, conversion stages 100 and entropy coder 102 are pressed
It is connected to according to the sequential series referred between the input 112 of Fig. 4 encoder and output end 114.The inverse transformation stage 104, add
Musical instruments used in a Buddhist or Taoist mass 110 and fallout predictor 106 according to being linked in sequence of referring to conversion stages 100 output end and subtracter 108 it is anti-phase defeated
Between entering end, the output end of fallout predictor 106 is also connected with another input of adder 110.
Fig. 4 encoder is the block coder based on predictive transformation.That is, array of samples 20 enters input
The block at end 112 is according to the previous coding of identical array of samples 20 and rebuilds part or other of previous coding or reconstruction sample
Array is predicted, and the presentation time of other array of samples can be before or after current array of samples 20.The prediction
Carried out by fallout predictor 106.Subtracter 108 subtracts prediction from such original block, and conversion stages 100 are entered to prediction residual
Row two-dimensional transform.The two-dimensional transform of the inner side of conversion stages 100 is in itself or follow-up measurement may result in conversion coefficient block
Conversion coefficient quantization.The conversion coefficient block of quantization for example carries out lossless volume by the entropy code in entropy coder 102
Code, and export resulting data flow in output end 114.The inverse transformation stage 104 rebuild quantify residual error, adder 110 by
The residual error of reconstruction is combined to obtain the message sample rebuild by this with corresponding prediction, and fallout predictor 106 is based on the reconstruction
Message sample can predict the prediction block of above-mentioned present encoding.Fallout predictor 106 can use different prediction moulds
Formula such as intra prediction mode and inter-frame forecast mode so as to prediction block and by Prediction Parameters be forwarded to entropy coder 102 with
Just data flow is inserted.For the prediction block of each inter prediction, each exercise data inserts ratio via entropy coder 114
To enable decoding side to give a forecast again in spy's stream.The exercise data of the prediction block of picture can include Grammar section, should
Grammar section include represent motion vector difference syntactic element, motion vector difference relative to for example by defined method from
Motion of the motion vector predictor that the motion vector of neighbouring encoded prediction block is drawn to current predictive block
Vector carries out differential coding.
That is, according to Fig. 4 embodiment, conversion coefficient block table show the spectral representation of the residual error of array of samples without
It is its actual message sample.That is, according to Fig. 4 embodiment, syntactic element sequence can enter entropy coder 102 with
Just entropy code is data flow 114.For transform blockiis, syntactic element sequence can include the motion for inter prediction block
Phasor difference syntactic element and be related to the position for indicating important transformation series several levels significance map (significance map) language
Method element and the syntactic element for the important transformation series several levels of transform blockiis restriction in itself.
It should be noted that Fig. 4 embodiment has some alternatives, a portion introduces portion the specification
Have been carried out describing in point, therefore the description is incorporated to Fig. 4 description.
Fig. 5 shows the decoder that can decode that the data flow by Fig. 4 encoder generation.Fig. 5 decoder includes entropy
Decoder 150, inverse transformation stage 152, adder 154 and fallout predictor 156.Entropy decoder 150, the inverse transformation stage 152 and add
Musical instruments used in a Buddhist or Taoist mass 154 is connected to according to the sequential series referred between the input 158 of Fig. 5 decoder and output end 160.Entropy decoding
Another output end of device 150 is connected with fallout predictor 156, the fallout predictor 156 be thus connected output end in adder 154 with
Between its another input.Entropy decoder 150 extracts conversion from the data flow of decoder for entering Fig. 5 in input 158
Coefficient blocks, wherein inverse transformation put on the conversion coefficient block in the stage 152 to obtain residual signal.Residual signal exists
Combined at adder 154 with the prediction from fallout predictor 156 to obtain the reconstructed version of array of samples in output end 160
Rebuild block.Based on reconstructed version, the generation prediction of fallout predictor 156, thus rebuild what is carried out by the fallout predictor 106 of coder side
Prediction.In order to which the prediction identical for obtaining with being used in coder side predicts that fallout predictor 156 is same using entropy decoder 150
In the Prediction Parameters that input 158 obtains from data flow.
It should be noted that in the above-described embodiments, the spatial granularity for carrying out prediction and the conversion of residual error need not be equal.
Such case is shown in Fig. 2 C.The figure has been shown in solid lines the subdivision of the prediction block of prediction granularity and illustrate with dashed lines
The subdivision of residual error granularity.As can be seen that subdivision can be selected by encoder independently from each other.More precisely, data flow grammer
It can allow to be defined residual error subdivision independently of prediction subdivision.Alternatively, residual error subdivision can be the expansion of prediction subdivision
Exhibition so that each residual error block is equal to the proper subclass of prediction block either prediction block.Shown on Fig. 2A and Fig. 2 B
Such case, for example, showing to predict granularity again wherein with solid line, residual error granularity is shown again with dotted line.That is,
In Fig. 2A -2C, all blocks with reference number related to this can carry out the residual error of a two-dimensional transform to it
Block, while the bigger solid line block comprising dotted line block 40a is, for example, the prediction that parameter setting is individually predicted to it
Block.
Above-described embodiment has in common that the block of (residual error is original) sample is transformed to convert in coder side
Thus coefficient blocks, the conversion coefficient block are inversely transformed into the reconstruction block of sample in decoder-side.Figure 6 illustrates
Such case.Fig. 6 shows the block of sample 200.For Fig. 6, the block 200 is exemplarily quadratic power and size
For 4x4 sample 202.Typically horizontally x and vertical direction y is set sample 202.Pass through above-mentioned two-dimensional transform T, block
200 are transformed to spectral domain, i.e. are transformed to the block 204 of conversion coefficient 206, size and the phase of block 200 of transform blockiis 204
Together.That is, transform blockiis 204 in the horizontal direction and the vertical direction with conversion coefficient 206 and block 200 with
Sample as many.However, because conversion T is spectral transformation, position of the conversion coefficient 206 in transform blockiis 204 not with space
Position correspondence, and it is corresponding with the spectral component of the content of block 200.Specifically, the trunnion axis of transform blockiis 204 corresponds to water
The axle that square upward spectral frequency is increased monotonically along it, and the spectral frequency that vertical axis corresponds in vertical direction is along its monotone increasing
The axle added, wherein DC components conversion coefficient are positioned in the turning (being illustratively the upper left corner here) of block 204 so that
The conversion coefficient 206 for corresponding to highest frequency in the horizontal direction and the vertical direction is located at the lower right corner.Ignore space side
To the spatial frequency belonging to some conversion coefficient 206 generally increases to the lower right corner from the upper left corner.Pass through inverse transformation T-1, transform blockiis
204 are transferred to spatial domain from spectral domain, to regain the copy 208 of block 200.During conversion not yet introduce quantify/
In the case of loss, reconstruction will be complete.
As has already been mentioned above, from fig. 6 it can be seen that the bigger resource block size of block 200 make it is resulting
The spectral resolution increase of spectral representation 204.On the other hand, quantizing noise is often distributed on whole block 208, therefore block 200
Interior mutation and the object that the localizes very much block that frequently can lead to remap made an uproar relative to original block 200 due to quantifying
Sound and deviation occurs.However, it is on the one hand important (i.e. non-zero (quantization)) conversion coefficient using the major advantage of bigger block
Quantity (i.e. level) and the bigger block of ratio on the other hand between the quantity of unessential conversion coefficient in can reduce
(compared with less block), thus improves code efficiency.In other words, frequently, important transformation series several levels (do not quantify
The conversion coefficient for being zero) sparse distribution is in transform blockiis 204.Due to this point, according to the implementation being described more fully below
Example, the position of important transformation series several levels carry out signal notice in data flow by significance map.Independently of this, important transformation series
Several value, i.e., the transformation series several levels when conversion coefficient is quantized, is transmitted in data flow.
Therefore all encoder and decoder as described above are configured to handle some grammer of syntactic element.Namely
Say, it is assumed that preceding syntax elements (such as transformation series several levels, be related to transform blockiis significance map syntactic element, to be related to interframe pre-
Survey exercise data syntactic element of block etc.) it is sequentially arranged in a prescribed manner in data flow.Mode as defined in this can
Represented in the form of with the pseudo-code as being presented, such as represented by H.264 standard or other coding and decoding videos.
Even in other words, foregoing description according to the rules some syntax element types, it is semantic and order therebetween pre-
Syntactic structure is defined, media data (being illustratively video data here) is converted to syntactic element sequence by main processing.Figure
4 and Fig. 5 entropy coder and entropy decoder is configurable to be operated, and may be configured, as next being summarized
's.It is also responsible in syntactic element sequence and data flow, i.e., being changed between symbol or bit stream.
Fig. 7 shows the entropy coder according to embodiment.The encoder turns the crossfire of syntactic element 301 without loss
Change one group of more than two partial bit stream 312 into.
In a preferred embodiment of the invention, the classification of each syntactic element 301 and one group of one or more classification is (i.e.
Syntax element type) it is associated.As example, the classification may specify the type of syntactic element.In the upper of hybrid video coding
Hereinafter, single classification can be with code macroblocks pattern, block coding pattern, reference picture index, motion vector difference, thin
Minute mark knowledge, encoded block mark, quantization parameter, transformation series several levels etc. are associated.Other application field such as audio, voice,
Text, file or conventional data coding, the classification of different syntactic elements is possible.
- as, each syntactic element can have limited value or countably infinite a class value, wherein, the group may
Syntax element value can be different for different syntactic element classifications.For example, binary syntax element and integer value be present
Syntactic element.
In order to reduce the complexity of coding and decoding algorithm, and in order to allow to different syntactic element and syntactic element
Classification carries out in general coding and decoding design, and syntactic element 301 is converted into orderly binary system decision set, then these
Binary system judges to pass through simple binary coding algorithm process.Therefore, binarizer 302 is by each syntactic element 301
It is worth bijective map to the sequence (or a string or a word) of bit (bin) 303.The sequence of the binary digit 303 represents
One group of orderly binary system judges.Each binary digit 303 or binary system judgement can obtain a value in two value sets,
For example, one in value 0 and 1.Binarization scheme can be different for different syntactic element classifications.For specific
Syntactic element classification, binarization scheme may depend on the set of possible syntax element value and/or the grammer of the particular category
Other properties of element.
Table 1 shows three example binary schemes for countably infinite set.For countably infinite set
Binarization scheme can also be applied to the finite aggregate of syntax element value.Especially for the finite aggregate of big syntax element value
Close, inefficiency can be ignored (caused by useless binary bit sequence), but such binarization scheme is logical
With property advantage is provided in terms of complexity and storage demand.It is often excellent for the finite aggregate of small syntax element value
Binarization project setting is adapted to the quantity of possible value of symbol by selection of land (for code efficiency).
Table 2 shows three example binary schemes of the finite aggregate for 8 values.With binary bit sequence
Finite aggregate represent the mode of irredundant code (and possible rearrangement to binary bit sequence), if passing through modification
Dry binary bit sequence, two for finite aggregate can be obtained from the general binary scheme for countably infinite set
Enter inhibition and generation scheme.As an example, the unary binarizations scheme of blocking in table 2 is entered by changing for general unitary two
The binary bit sequence of the syntactic element 7 of inhibition and generation and form (with reference to table 1).The block and rearrangement of order 0 in table 2
Index Columbus binary system pass through modification be used for general-purpose indexes Columbus order 0 binary syntactic element 7 binary digit
Sequence (being shown in Table 1) and by binary bit sequence of resequencing (be used for symbol 7 the two-stage system bit sequence that blocks be assigned to symbol
Number 1) formed.For the finite aggregate of syntactic element, non-systemic/non-universal property binarization scheme, such as table can be used
Shown in 2 last column.
Table 1：Binary instance for countably infinite set (or big finite aggregate).
Table 2：Binary instance for finite aggregate.
Each binary digit (bin) 303 of the binary bit sequence formed by binarizer 302 is by order feed
Parametric distribution device 304.The set of one or more parameters is distributed to each binary digit 303 by parametric distribution device, and exports tool
There is the binary digit that associated parameter 305 is gathered.Parameter sets are true in a manner of identical at encoder and decoder
It is fixed.Parameter sets can be made up of one or more of following parameters:
Specifically, parametric distribution device 304 is configurable to distribute context model to current binary digit 303.For example,
Parametric distribution device 304 can be directed to one in current binary digit 303 selection available contexts index.For when the first two enters
The available contexts set of position 303 processed can depend on the type of binary digit, and the type of the binary digit is on the contrary can be by language
The type/genre of method element 301 limits, and a binary part for the syntactic element 301 is current binary digit 303, and
It is the position of current binary digit 303 in the binary system of the latter.Context selection in available contexts set can take
Certainly in previous binary digit and the syntactic element associated with the latter.Each of these contexts has associated therewith
Probabilistic model (that is, the measurement of the probability Estimation of one of two possible binary bit values for current binary digit of connection
Value).Probabilistic model can be specifically to be directed to carry out for the low probability or high probability binary bit value of current binary digit
The measured value of probability Estimation, wherein probabilistic model are in addition by specifying which of two possible binary bit values to represent current
The identifier of the estimation of the low probability or high probability binary bit value of binary digit 303 limits.To current binary system
In the case of an only context is available for position, it is convenient to omit context selects.Following article is summarized in more detail, parameter
Distributor 304 can also carry out probabilistic model and adapt to make to adapt to belong to each with each context-sensitive probabilistic model
The actual binary position statistics of each binary digit of context.
Equally as described in more detail below, parametric distribution device 304 can be according to being activation high efficiency (HE) pattern or low
Complexity (LC) pattern and carry out different operating.Under the two patterns, probabilistic model makes current binary digit 303 enter with two
Any one of position encoder 310 processed is associated, as outlined below, but the operator scheme of parametric distribution device 304 is in LC moulds
Tend under formula it is less complicated, however, code efficiency under high efficiency mode because parametric distribution device 304 makes independent two to enter
Position 303 processed is associated with individual encoders 310 to be improved with more accurately adapting to binary digit statistics, thus relative to LC moulds
Formula optimizes entropy.
Each binary digit that the parameter 305 that tool as the output of parametric distribution device 304 is associated is gathered is fed into two and entered
Digit buffer selector 306 processed.Binary digit buffer selector 306 is based on input binary value and associated parameter 305
Potentially to change the value of input binary digit 305, and by the feed-in of output binary digit 307 with the value potentially changed
In one in two or more binary buffers 308.Output binary digit 307 is sent to binary digit buffering therein
Device 308 is determined based on the value of input binary digit 305 and/or the value of associated parameter 305.
In a preferred embodiment of the invention, binary digit buffer selector 306 and unmodified binary bit value, i.e.
Exporting binary digit 307 generally has with inputting the identical value of binary digit 305.In the another preferred embodiment of the present invention,
Two possible binary digits of the binary digit buffer selector 306 based on input binary bit value 305 and current binary digit
The associated measured value of the probabilistic estimated value of one in value come determine export binary bit value 307.In the preferred of the present invention
In embodiment, if currently the measured value of the probability of one in two possible binary bit values of binary digit is less than (or
Less than or equal to) certain threshold, then export binary bit value 307 and be set equal to input binary bit value 305；If
The probability measure of one in two possible binary bit values of current binary digit is more than or equal to that (or more than) is specific to be faced
Dividing value, then export binary bit value 307 and changed (that is, it is set to opposite with input binary bit value).The present invention's
In another preferred embodiment, if the probability measure of one in two possible binary bit values of current binary digit is big
In (or more than or equal to) certain threshold, then export binary bit value 307 and be set equal to input binary bit value 305；
If currently the probability measure of one in two possible binary bit values of binary digit is less than or equal to (or less than) special
Determine critical value, then export binary bit value 307 and changed (that is, it is set to opposite with input binary bit value).In this hair
In bright preferred embodiment, for the estimated probability of two possible binary bit values, critical value corresponds to value 0.5.
In the another preferred embodiment of the present invention, binary digit buffer selector 306 is based on input binary bit value
Which of 305 and two possible binary bit values of instruction represent the low probability or high probability two of current binary digit
The associated identifier of the estimate of system place value come determine export binary bit value 307.In the preferred embodiments of the present invention
In, if identifier specifies first in two possible binary bit values to represent that current binary digit low probability is (or higher
Probability) binary bit value, then export binary bit value 307 and be set equal to input binary bit value 305；And if mark
Know the low probability (or high probability) two that symbol specifies second current binary digit of expression in two possible binary bit values
System place value, then export binary bit value 307 and changed (that is, it is set to opposite with input binary bit value).
In a preferred embodiment of the invention, two based on current binary digit of binary digit buffer selector 306
The associated measured value of the probabilistic estimated value of one in possible binary bit value is sent out to determine to export binary digit 307
The binary digit buffer 308 delivered to.In a preferred embodiment of the invention, one in two possible binary bit values
The collection of the probable value of the measured value of probabilistic estimated value is combined into limited, and binary digit buffer selector 306 includes a table, should
Table is each by lucky bit buffer 308 and the probabilistic estimated value of one in two possible binary bit values
Probable value is associated, wherein, the different value of the probabilistic estimated value measured value of one in two possible binary bit values can with it is same
Bit buffer 308 is associated.In the another preferred embodiment of the present invention, in two possible binary bit values
The probable value scope of measured value of the probabilistic estimated value of one be divided into multiple sections, binary digit buffer selector
306 determine the section index of the current measurement value of the probabilistic estimated value of one in two possible binary bit values, and binary system
Digit buffer selector 306 includes a table, and the table indexes lucky bit buffer 308 with the section each
Probable value is associated, and the different value of wherein section index can be associated with same binary digit buffer 308.In the present invention
Preferred embodiment in, have the opposite measured value of the probabilistic estimated value of one in two possible binary bit values (opposite to survey
Value is represents probabilistic estimated value P and 1-P measured value) input binary digit 305 be fed into same binary digit buffering
Device 308.In the another preferred embodiment of the present invention, one in two possible binary bit values of current binary digit
The correlation of the measured value of probabilistic estimated value and specific binary digit buffer is adapted over time, for example, to ensure
The partial bit stream of formation has similar bit rate.Further below, section index will be referred to as pipeline index, manage simultaneously
Road indexes gives actual probabilities model (that is, probability Estimation together with the mark of refinement index and instruction high probability binary bit value
Value) index.
In the another preferred embodiment of the present invention, binary digit buffer selector 306 is based on current binary digit
The correlation measure of the probabilistic estimated value of low probability or high probability binary bit value and determine export the quilt of binary digit 307
Send binary digit buffer 308 extremely.In a preferred embodiment of the invention, low probability or high probability binary digit
The collection of the probable value of the measured value of the probabilistic estimated value of value be combined into it is limited, binary digit buffer selector 306 include one
Table, the table is just by bit buffer 308 and the probabilistic estimated value of low probability or high probability binary bit value
Each probable value be associated, wherein, the difference of the probabilistic estimated value measured value of low probability or high probability binary bit value
Value can be associated with same binary digit buffer 308.The present invention another preferred embodiment in, low probability or compared with
The probable value scope of the probabilistic estimated value measured value of high probability binary bit value is divided into multiple sections, binary digit buffering
Device selector 306 determines the section index of the current probability estimate measured value of low probability or high probability binary bit value,
Binary digit buffer selector 306 includes a table, what the table just indexed bit buffer 308 and section
Each probable value is associated, wherein, the different value of section index can be associated with same binary digit buffer 308.At this
In the another preferred embodiment of invention, the low probability of current binary digit or the probability Estimation of high probability binary bit value
The correlation of value measured value and specific binary digit buffer is adapted over time, such as to ensure formed part
Bit stream has similar bit rate.
Each of two or more binary digit buffers 308 is just connected with bit encoder 310, and
Each binary digit encoder is only connected with bit buffer 308.Each binary digit encoder 310 is from associated
Binary digit buffer 308 read binary digit, and by binary bit sequence 309 be converted into represent bit sequence code word
311.Binary digit buffer 308 represents first-in first-out buffer；Binary digit buffer is relatively fed into (sequentially) late
308 binary digit will not be compiled before in the relatively early binary digit that the binary digit buffer is fed into (sequentially)
Code.Code word 311 as the output of specific binary digit encoder 310 is written to specific part bit stream 312.It is whole to compile
Syntactic element 301 is converted to two or more partial bit streams 312 by code algorithm, wherein, the quantity of partial bit stream is equal to two
The quantity of system digit buffer and binary digit encoder.In a preferred embodiment of the invention, binary digit encoder 310
The binary digit 309 of variable number is converted into the code word 311 of variable amount of bits.The embodiment of the present invention above-mentioned and with shown below
The advantage gone out is that binary digit coding can carry out (for example, for different probability measure groups) side by side, this reduction
Processing times of some specific implementations.
Another advantage of the embodiment of the present invention is can be that different parameters 305 gather specific design by encoder
The 310 binary digit codings performed.Specifically, the probability group for different estimations can optimize (with regard to code efficiency and/or again
For polygamy) binary digit encodes and coding.On the one hand, this allows to reduce coding/decoding complexity.On the other hand, this
Allow to improve code efficiency.In a preferred embodiment of the invention, binary digit encoder 310 is different groups and works as the first two
The measured value of the probabilistic estimated value of one in two possible binary bit values 305 of system position performs different encryption algorithms
(that is, binary sequence maps to code word).In the another preferred embodiment of the present invention, binary digit encoder 310 is not
The measured value of the low probability of the current binary digit of same group or the probabilistic estimated value of high probability binary bit value performs difference
Encryption algorithm.
In the another preferred embodiment of the present invention, one in binary digit encoder 310 or binary digit encoder
Or multiple entropy coders for representing to map directly to input binary bit sequence 309 into code word 310.Such mapping relations
It can effectively realize and not require complicated arithmetic coding engine.Inverse mapping of the code word to binary bit sequence (is such as decoding
Conducted in device) must be uniquely to ensure that the perfect of list entries decodes, but binary bit sequence 309 is to code word 310
Mapping be not necessarily unique, i.e. specific binary bit sequence can map to more than one codeword sequence.
In the preferred embodiments of the present invention, binary bit sequence 309 to code word 310 is mapped as dijection.In the another of the present invention
In preferred embodiment, one or more of binary digit encoder 310 or binary digit encoder are represented variable-length
Input binary bit sequence 309 maps directly to the entropy coder of the code word 310 of variable-length.In the preferred reality of the present invention
Apply in example, output code word represents irredundant code, such as the Huffman code of in general Huffman code or standard.
Show binary bit sequence to two examples of the bijective map of irredundant code in table 3.The present invention's
In another preferred embodiment, output code word represents to be suitable for the redundant code of error detection and Fault recovery.In the present invention
Another preferred embodiment in, output code word represents to be suitable for the encrypted code of Encrypt the Grammar element.
Table 3：The example of mapping between binary bit sequence and code word.
In the another preferred embodiment of the present invention, one in binary digit encoder 310 or binary digit encoder
Or the entropy of multiple code words 310 for representing the input binary bit sequence 309 of variable-length mapping directly to regular length is compiled
Code device.In the another preferred embodiment of the present invention, one in binary digit encoder 310 or binary digit encoder or
Multiple entropy codes for representing to map directly to the input binary bit sequence 309 of regular length into the code word 310 of variable-length
Device.
Fig. 8 shows decoder according to embodiments of the present invention.Decoder essentially performs the inverse operation of encoder, from
And the set of (previously encoded) syntactic element sequence 327 from two or more partial bit streams 324 is decoded.Decoder
Including two different processing streams：One is flowed for request of data, the data flow of its replica code device；And data flow, represent to compile
The inverse operation of code device data flow.As shown in Figure 8, dashed arrows show data request stream, and solid arrow represents data flow.
The structure module of the structure module of decoder substantially replica code device, but perform inverse operation.
The decoding of syntactic element is touched by being sent to the new request 313 for decoding syntactic element of binarizer 314
Hair.In a preferred embodiment of the invention, each and one or more classifications of the request 313 of new decoding syntactic element
Set in classification be associated.The classification associated with the request of syntactic element is first with corresponding grammer with during coding
The associated classification of element is identical.
Binarizer 314 maps to the request 313 of syntactic element the binary digit for being sent to parametric distribution device 316
One or more requests.Request as binary digit that parametric distribution device 316 is sent to by binarizer 314
Finally respond, binarizer 314 receives the binary digit 326 of the decoding from binary digit buffer selector 318.Two
Enter specific binarization of the inhibition and generation device 314 by the sequence of the binary digit 326 of the decoding received and the syntactic element asked
The binary bit sequence of scheme is compared, if the sequence of the binary digit 326 of the decoding received and syntactic element
Binarization matches, then binarizer empties its binary system digit buffer and as the final response to new solution code sign
And export the syntactic element of decoding.If the binary bit sequence of the decoding received the two of syntactic element enters with what is asked
The binary bit sequence of inhibition and generation scheme mismatches, then binarizer sends the request of another binary digit to parametric distribution device,
One in the binary bit sequence of decoding and the binary bit sequence for the binarization scheme of syntactic element asked
Untill matching.Asked for the items of syntactic element, decoder, which uses, is used for what corresponding syntactic element was encoded
Identical binarization scheme.For different syntactic element classifications, binarization scheme may be different.Specific syntax elements class
Other binarization scheme may depend on other properties of possible syntactic element value set and/or particular category syntactic element.
The set of one or more parameters is distributed to each request of binary digit by parametric distribution device 316, and will be had
The request of the binary digit of associated parameter sets is sent to binary digit buffer selector.Pass through parametric distribution device point
The parameter sets of the parameter sets for the binary digit that dispensing is asked binary digit corresponding with being assigned to during coding are
Identical.Parameter sets can be made up of one or more of parameter mentioned in the encoder description in Fig. 7.
In a preferred embodiment of the invention, every request and distributor 304 of the parametric distribution device 316 by binary digit
The identical parameters distributed (that is, one in two possible binary bit values of context and currently requested binary digit
Probabilistic estimated value associated measurement value, for example the low probability of currently requested binary digit or high probability two enter
The measured value of the probabilistic estimated value of place value processed, and specify which of two possible binary bit values to represent currently requested
Binary digit low probability or high probability binary bit value estimate identifier) it is associated.
Parametric distribution device 316 can determine above-mentioned probability measurement based on the set of one or more decoded symbols
One or more of the value (probability Estimation of one in two possible binary bit values of currently requested binary digit
The probabilistic estimated value of the measured value of value, the low probability of currently requested binary digit or high probability binary bit value
Measured value, specify which of two possible binary bit values represent currently requested binary digit low probability or
The identifier of the estimate of high probability binary bit value).Determination for the probability measure of the specific request of binary digit
Replicate the processing at the encoder for corresponding binary digit.The symbol of decoding for determining probability measure may include
The symbols of one or more decodings of same-sign classification, corresponding to neighbouring space and/or time location (and with grammer member
The data acquisition system that the current request of element is associated is related) data acquisition system (such as block or sample group) same-sign classification
One or more decoded symbols or corresponding to identical and/or neighbouring space and/or time location (and with grammer member
The data acquisition system that the current request of element is associated is related) the one or more of distinct symbols classification of data acquisition system decoded
Symbol.
Every request as the binary digit with associated parameter sets 317 of the output of parametric distribution device 316
It is fed into binary digit buffer selector 318.Based on associated parameter sets 317, binary digit buffer selector
318 send the request 319 of binary digit to one in two or more binary digit buffers 320, and from selected two
System digit buffer 320 receives the binary digit 325 of decoding.The input binary digit 325 of decoding is potentially changed, and is had
There is the output binary digit 326 of the decoding of potential modified values as the binary digit with associated parameter sets 317
The final response of request is sent to binarizer 314.
With with the output binary digit in coder side binary digit buffer selector transmitted by binary digit extremely delay
Rush the binary digit buffer 320 that device identical mode selects the request of binary digit to be forwarded to.
In a preferred embodiment of the invention, binary digit buffer selector 318 is based on currently requested binary system
The correlation measure of the probabilistic estimated value of one in two possible binary bit values of position, determines the request 319 of binary digit
Transmitted binary digit buffer 320 extremely.In a preferred embodiment of the invention, one in two possible binary bit values
The collection of the probable value of the measured value of individual probabilistic estimated value be combined into it is limited, and binary digit buffer selector 318 include it is proper
Good each possibility by the probabilistic estimated value of one in bit buffer 320 and two possible binary bit values
The associated table of value, wherein, the different value of the measured value of the probabilistic estimated value of one in two possible binary bit values can be with
Same binary digit buffer 320 is associated.In the another preferred embodiment of the present invention, two possible binary bit values
In the probable value scope of measured value of the probabilistic estimated value of one be divided into multiple sections, the selection of binary digit buffer
Device 318 determines the section index of the current measurement value of the probabilistic estimated value of one in two possible binary bit values, binary system
Digit buffer selector 318 includes just that bit buffer 320 is related to each probable value that section indexes
The table of connection, wherein, the different value of section index can be associated with same binary digit buffer 320.In the preferred of the present invention
In embodiment, there is opposite measured value (the opposite measured value table of the probabilistic estimated value of one in two possible binary bit values
Show probabilistic estimated value P and 1-P those measured values) the request 317 of binary digit be forwarded to same binary digit buffering
Device 320.In the another preferred embodiment of the present invention, one in two possible binary bit values of current binary digit request
The relevance of the measured value of individual probabilistic estimated value and a specific binary digit buffer is adapted over time.
In the another preferred embodiment of the present invention, binary digit buffer selector 318 is based on currently requested two
The low probability of system position or the correlation measure of the probabilistic estimated value of high probability binary bit value determine binary digit
Binary digit buffer 320 transmitted by request 319 extremely.In a preferred embodiment of the invention, low probability or high probability
The collection of the probable value of the measured value of the probabilistic estimated value of binary bit value is combined into limited, and binary digit buffer selector
318 comprising just by bit buffer 320 and the probability Estimation of low probability or high probability binary bit value
The associated table of each probable value of value, wherein, the measurement of the probabilistic estimated value of low probability or high probability binary bit value
The different value of value can be associated with same binary digit buffer 320.It is relatively low in the another preferred embodiment of the present invention
The scope of the probable value of the measured value of probability or the probabilistic estimated value of high probability binary bit value is divided into multiple sections,
Binary digit buffer selector 318 determines the current survey of the probabilistic estimated value of low probability or high probability binary bit value
The section index of value, and binary digit buffer selector 318 are included just by bit buffer 320 and area
Between the associated table of each probable value for indexing, wherein, the different value of section index can be with same binary digit buffer
320 is associated.In the another preferred embodiment of the present invention, the low probability or high probability two of current binary digit request
The relevance of the measured value of the probabilistic estimated value of system place value and specific binary digit buffer is adapted over time.
After the binary digit 325 of decoding is received from the selected binary digit buffer 320, binary digit buffering
Device selector 318 potentially changes input binary digit 325 and sends the output binary digit 326 with potential modified values
To binarizer 314.The input/output binary digit of binary digit buffer selector 318 is mapped as at coder side
Binary digit buffer selector input/output binary digit mapping inverse operation.
In a preferred embodiment of the invention, binary digit buffer selector 318 and unmodified binary bit value, i.e.
Exporting binary digit 326 always has with inputting the identical value of binary digit 325.In the another preferred embodiment of the present invention,
Binary digit buffer selector 318 based on input binary bit value 325 and it is associated with the request 317 of binary digit ought
The measured value of the probabilistic estimated value of one in two possible binary bit values of preceding asked binary digit and determine to export
Binary bit value 326.In a preferred embodiment of the invention, if two possible binary digits of current binary digit request
The probability measure of one in value is less than (or less than or equal to) certain threshold, then exports binary bit value 326 and be set
For equal to input binary bit value 325；If one 's in two possible binary bit values of current binary digit request is general
Rate measured value is more than or equal to (or more than) certain threshold, then export binary bit value 326 changed (that is, be set as with it is defeated
It is opposite to enter binary bit value).In the another preferred embodiment of the present invention, if two possibility of current binary digit request
The probability measure of one in binary bit value is more than (or more than or equal to) certain threshold, then exports binary bit value
326 are set equal to input binary bit value 325；If in two possible binary bit values of current binary digit request
The probability measure of one be less than or equal to (or less than) certain threshold, then export binary bit value 326 and changed (i.e.,
It is set as opposite with input binary bit value).In a preferred embodiment of the invention, critical value is entered corresponding to two possible two
0.5 value of the estimated probability of place value processed.
In the another preferred embodiment of the present invention, binary digit buffer selector 318 is based on input binary bit value
325 and the identifier associated with the request 317 of binary digit determine to export binary bit value 326, the identifier specifies
Which of two possible binary bit values represent the estimation of the low probability or high probability of current binary digit request.
In a preferred embodiment of the invention, if identifier specifies first expression in two possible binary bit values to work as the first two
Low probability (or high probability) binary bit value of system position request, then export binary bit value 326 and be set equal to
Input binary bit value 325；And if identifier specifies second in two possible binary bit values to represent current binary system
Position request low probability (or high probability) binary bit value, then export binary bit value 326 change (that is, be set as and
It is opposite to input binary bit value).
As described above, the request 319 of binary digit is sent to two or more two by binary digit buffer selector
One in digit buffer 320 processed.Binary digit buffer 320 represents first-in first-out buffer, and it is presented with from connection
The binary sequence 321 of the decoding of binary digit decoder 322.As to from binary digit buffer selector 318 send to
The response of the request 319 of the binary digit of binary digit buffer 320, binary digit buffer 320 remove first feed binary system
The binary digit of the content of digit buffer 320, and send it to binary digit buffer selector 318.Relatively early hair
The binary digit for giving binary digit buffer 320 is removed and is sent to binary digit buffer selector earlier
318。
Each in two or more binary digit buffers 320 is just connected with bit decoder 322,
And each binary digit decoder is only connected with bit buffer 320.Each binary digit decoder 322 from point
The partial bit stream 324 opened reads the code word 323 for representing bit sequence.Code word 323 is converted into sending out by binary digit decoder
Deliver to the binary bit sequence 321 of the binary digit buffer 320 of connection.Whole decoding algorithm is by two or more part ratios
Spy's stream 324 is converted into the syntactic element of multiple decodings, wherein, the quantity of partial bit stream is equal to binary digit buffer and two
The quantity of system position decoder, and the decoding of syntactic element is triggered by the request of new syntactic element.In the excellent of the present invention
Select in embodiment, it is variable that the variable code word 323 of bit quantity is converted into binary digit quantity by binary digit decoder 322
Sequence 321.One advantage of embodiments of the invention is, is decoded from the binary digit of two or more partial bit streams
(for example, different groups for probability measure) it can carry out side by side, which reduce the processing time of some specific implementations.
Another advantage of embodiments of the invention is, the binary digit carried out by binary digit decoder 322
Decoding can specific design for different parameters set 317.Specifically, for the probability group of different estimations, can optimize
(for code efficiency and/or complexity) binary digit encoding and decoding.On the one hand, this allows relative to prior art
Entropy code algorithm with similar code efficiency can reduce coding/decoding complexity.On the other hand, this allows relative to existing
The entropy code algorithm with similar coding/decoding complexity for having technology can improve code efficiency.In the preferred of the present invention
In embodiment, one in two possible binary bit values 317 that binary digit decoder 322 is asked for current binary digit
The different groups of the measured value of individual probabilistic estimated value perform different decoding algorithm, and (that is, binary sequence to code word reflects
Penetrate).The present invention another preferred embodiment in, binary digit decoder 322 for currently requested binary digit compared with
The different groups of the measured value of the probabilistic estimated value of low probability or high probability binary bit value perform different decoding algorithms.
Binary digit decoder 322 performs the inverse mapping of the corresponding binary digit encoder at encoder.
In a preferred embodiment of the invention, one or more in binary digit decoder 322 or binary digit decoder
The individual entropy decoder for representing to map directly to code word 323 into binary bit sequence 321.Such mapping can be by effectively real
Now without requiring complicated arithmetic coding engine.The mapping of codeword mappings to binary sequence must be unique.In this hair
In bright preferred embodiment, code word 323 maps to the dijection that is mapped as of binary bit sequence 321.In the another of the present invention
In preferred embodiment, one or more of binary digit decoder 310 or binary digit decoder are represented variable-length
Code word 323 map directly to variable-length binary bit sequence 321 entropy decoder.In being preferable to carry out for the present invention
In example, input code word represents irredundant code, such as the Huffman code of in general Huffman code or standard.Shown in table 3
Two examples for the bijective map of irredundant code to binary bit sequence.
In the another preferred embodiment of the present invention, one in binary digit decoder 322 or binary digit decoder
Or multiple entropy decoders for representing to map directly to the code word 323 of regular length into the binary sequence 321 of variable-length.
In the another preferred embodiment of the present invention, one or more of binary digit decoder 322 or binary digit decoder table
Show the entropy decoder for the binary sequence 321 that the code word 323 of variable-length is mapped directly to regular length.
Therefore, Fig. 7 and Fig. 8 shows for the encoder of coded identification sequence 3 and for rebuilding the decoding of symbol sebolic addressing
The embodiment of device.Encoder includes the distributor for each symbol for being configured as distributing to multiple parameters 305 symbol sebolic addressing
304.It is described distribution the information based on the prior symbol included in the symbol sebolic addressing, such as the classification of syntactic element 1 and distribute
To the expression (such as binarization) belonging to current sign, and the syntactic structure according to syntactic element 1, the expression is current
It is contemplated that the expection can estimate from the history of previous grammar element 1 and symbol 3 again.In addition, encoder is compiled comprising multiple entropys
Code device 10 and selector 306, each symbol 3 for being configured as will be forwarded to each entropy coder of the entropy coder are changed
Into each bit stream 312, selected by the selector is configured as each symbol 3 being forwarded in multiple entropy coders 10
One, it is described selection depend on distribute to each symbol 3 parameter 305 quantity.Distributor 304 can be considered as integrated
Into selector 206 to obtain each selector 502.
Decoder for rebuilding symbol sebolic addressing includes multiple entropy decoders 322, and each entropy decoder is configured as will be each
Individual bit stream 323 is converted into symbol 321；Distributor 316, it is configured as the symbol based on the previous reconstruction included in symbol sebolic addressing
Information (refer to the 8th figure 326 and 327) in number distributes to multiple parameters 317 by each symbol of reconstructed sequence symbol
Numbers 315；And selector 318, be configured as from multiple entropy decoders 322 selected one recover by be reconstructed
Each symbol of symbol sebolic addressing, the selection depend on limiting to the number of parameters of each symbol.Distributor 316 can be configured
Make it that it is general in the presumable multiple possible values of symbol of (either) each symbol that the number of parameters for distributing to each symbol includes
Rate is distributed the measured value of estimate.Furthermore distributor 316 and selector 318 can be considered as being integrated in a block, selection
In device 402.Can be binary alphabet by reconstructed symbol sebolic addressing, distributor 316 can be configured such that probability distribution
Estimate by the binary alphabet two possible binary bit values low probability or high probability binary bit value
Probabilistic estimated value measured value and specify which of two possible binary bit values to represent low probability or high probability
The identifier of the estimate of binary bit value is formed.Distributor 316 is further configurable to be based on being included in using each
By the symbol of the previous reconstruction of reconstructed symbol sebolic addressing, (each context has each probability associated there point to context
Cloth estimate) in information and context is internally assigned to each symbol of reconstructed symbol sebolic addressing 315；And base
The value of symbol that the symbol of previous reconstruction extremely is allocated in each context adjusts the Distribution estimation value of each context
Fit to actual symbol statistical value.The context is contemplated that (such as in video or Image Coding, or even the feelings in financial application
In the table of condition) spatial relationship belonging to syntactic element or position close on.Then, based on distributing to the upper of each symbol
Hereafter associated Distribution estimation value, such as by (will be indexed in the examples below with pipeline with each symbol of utilization
Indexed with together with refinement index) distribute to one during multiple Distribution estimation values represent and (remove refinement index) up and down
The associated Distribution estimation value of text quantifies or used as the index of each table, and determine the probability distribution of each symbol
The measured value of estimate, to obtain the measured value of the estimate of the probability distribution (the pipeline rope of indicating section bit stream 312
Draw).Selector can be configured such that defining dijection between multiple entropy coders and multiple Distribution estimation values represent closes
Connection property.Selector 18 can be configured to predetermined qualitative fashion really, and (which depends on the symbol of the previous reconstruction of the symbol sebolic addressing
Number), the quantization represented over time and by change from the scope of Distribution estimation value to multiple Distribution estimation values is reflected
Penetrate.That is, selector 318, which can change, quantifies step size and (that is, maps to each entropy decoder dijection associated each
The section of the probability distribution of probability index).Thus multiple entropy decoders 322 can be configured to respond to quantify the change of mapping
And adjust their mode that symbol is converted into bit stream.For example, each entropy decoder 322 can be directed in each probability
Some the Distribution estimation value being distributed in estimate quantized interval is optimised (i.e., it is possible to optimal compression speed)；
And its code word/symbol sebolic addressing mapping can be changed, so as to when each Distribution estimation value quantized interval changes, adjust
Position of some the Distribution estimation value in each Distribution estimation value quantized interval and it is optimised.Selector can quilt
It is configured to change and quantifies mapping so that the speed that symbol is resumed from multiple entropy decoders becomes less scattered.As for binary system
Change device 314, it should be noted that identical binarizer can be removed if syntactic element is binary system.This
Outside, according to the type of entropy decoder 322, the presence of buffer 320 is not necessary.In addition, buffer can be integrated in
In decoder.
The termination of limited syntactic element sequence
In a preferred embodiment of the invention, limited syntactic element set is coded and decoded.It is generally a certain amount of
Data (such as frame of still image, video sequence or field, the fragment of image, the fragment of the frame of video sequence or field,
Or the set of continuous audio sample) it is coding.Typically for limited syntactic element set, formed at coder side
Partial bit stream must be terminated, i.e. must assure that whole syntactic elements can be from partial bit that is being transmitted or being stored
Stream decoding.After binary digit buffer 308 corresponding to being inserted into last binary digit, binary digit encoder 310
It must assure that complete code word is written to partial bit stream 312.If binary digit encoder 310 represents to perform binary system
Bit sequence to the direct mapping of code word entropy coder, then by last binary digit be written to binary digit buffer it
Afterwards, being stored in the binary bit sequence of binary digit buffer may not represent that (that is, it may be represented and generation with code word
The prefix of the associated two or more binary bit sequences of code word) associated binary bit sequence.In such situation
Under, with including any generation associated as the binary bit sequence of the binary bit sequence of prefix in binary digit buffer
Code word must be written to partial bit stream (binary digit buffer must be refreshed).This can by will have particular value or
Carried out in the binary digit insertion binary digit buffer of arbitrary value untill code word is written into.In the excellent of the present invention
Select in embodiment, the selection of binary digit encoder is with minimum length (except associated binary bit sequence must include two
In system digit buffer as the property of the binary sequence of prefix outside) code word in one.In decoder-side, two
The decodable code of system position decoder 322 is than more binary digits required for last code word in partial bit stream；This
A little binary digits are not asked by binary digit buffer selector 318 so as to be rejected and ignore.Limited assemble of symbol
Decoding by asking the syntactic element of decoding to control；If no longer asking some data volume of syntactic element, solution is terminated
Code.
The transmission and multiplexing of partial bit stream
The partial bit stream 312 formed by encoder can be transmitted dividually, or can be demultiplexed into single bit
Stream, or the code word of partial bit stream can be interleaved in single bit stream.
In an embodiment of the present invention, each several part bit stream of the amount of some data is written to a packet.These numbers
According to amount can be syntactic element (such as the frame of static images, video sequence or field, the fragment of static images, video sequence
Frame or the fragment of field or the frame etc. of audio sample) arbitrary collection.
The present invention another preferred embodiment in, two or more partial bit streams of the amount of some data or some
The all parts bit stream of the amount of data is multiplexed into a packet.Figure 9 illustrates the portion comprising multiplexing
Divide the structure of the packet of bit stream.
Packet 400 is made up of a segmentation of header and the data of each several part bit stream (data volume considered).
The header 401 of packet includes the mark that packet (remainder) is divided into multiple bitstream data fragments 402.Remove
Segmentation mark is outer, and header can include other information.In a preferred embodiment of the invention, the segmentation of packet is identified as
The position of data slot starting point in units of position or byte or multiple positions or multiple bytes.In the preferred embodiments of the present invention
In, the position of data slot starting point is relative to packet starting point or relative to header terminal or relative to previous packet
Starting point is encoded as the absolute value in packet header.In the another preferred embodiment of the present invention, the position of data slot starting point
Put and differentially encoded, i.e. only the difference between the actual start of data slot and the prediction of data slot starting point is compiled
Code.Information (size of population of such as packet, header size, the data slice in packet that can be based on Given information or transmission
The quantity of section, the position of previous data slot starting point) and obtain the prediction.In a preferred embodiment of the invention,
The position of one packet starting point is not encoded, but is deduced based on packet header size.In decoder-side, passed
Defeated segmentation mark is used for obtaining data slot starting point.Then, data slot is used as partial bit stream, and included in data
Data in the fragment binary digit decoder corresponding to sequentially feed.
Some optional methods are surrounded by for partial bit stream is multiplexed into data.Figure 10 illustrates can subtract
A kind of alternative of few required side information (side information) is (especially for which part bit stream
The very much like situation of size).The payload (that is, the packet 410 in addition to its header 411) of packet is with predetermined
Mode and be divided into multiple fragments 412.As an example, the payload of packet can be divided into identical
The fragment of size.Then each fragment and partial bit stream or associated with the Part I of partial bit stream 413.Such as fruit part
Bit stream is more than associated data slot, then its remainder 414 is placed on the vacant space of other data slot ends
In.This can be carried out as follows：The remainder of bit stream is inserted with backward (end for starting from data slot), so as to
Reduce side information.The remainder of partial bit stream is associated with data slot, and when more than one remainder is added
During to data slot, the starting point in one or more remainders must be inside bit stream (for example, in packet header
In) signal notice.
The intertexture of the code word of variable-length
For some applications, the multichannel of the above-mentioned partial bit stream (for quantitative syntactic element) in a packet is answered
With can have the disadvantage that：On the one hand, for small-sized packet, for sending the position of the side information required by splitting signal
(bit) quantity may become notable relative to the real data in partial bit stream, and this eventually reduces code efficiency.It is another
Aspect, multiplexing may be not suitable for requiring the application (such as video conference application) of low latency.Answered by foregoing multichannel
With before partial bit stream is formed completely, encoder can not start the transmission of packet, and reason is to split starting point position
Put be not known before.In addition, in general, decoder must be waited until its data slot of reception to the end
Starting point, then can just start the decoding of packet.For the application as video conferencing system, these delays more increase system
Unite some video images overall delay (especially for the bit rate close to transmission bit rate, and for requiring between two images
Time interval close to for coding/decoding picture encoder/decoder), it is very fatal to such application.For
The shortcomings that overcoming some applications, the encoder of the preferred embodiments of the present invention can configure in the following manner, by two or more
Code word caused by individual binary digit encoder is woven into single bit stream.Bit stream with interleaved code word can be straight
Decoder (when ignoring small buffer delay, seeing below) is delivered in sending and receiving.In decoder-side, two or more binary digit solutions
Code device directly reads code word with decoding order from bit stream；Decoding may begin at the position of first reception.Furthermore, it is not required that side
Side information is with the signal of the multiplexing (or intertexture) for transmitting portion bit stream.When binary digit decoder 322 not from
General digit buffer reads variable length codewords, but they generally read fixed-length bits sequence from general digit buffer
Arrange and (wherein, each binary digit decoder 322 is with separating added to local digit buffer by these regular length bit sequences
Local digit buffer connection) when, can obtain reduce decoder complexity other mode.Then delay from this status
Rush device and read variable length codewords.Therefore, the anatomy of variable length codewords can be carried out side by side, only regular length position sequence
The access of row must be carried out in a synchronous manner, but the access of such regular length bit sequence is generally very quick, so as to right
In some structures, whole decoding complexity can be reduced.The binary digit quantity for being sent to the fixation of specific local digit buffer can
It is different because of different local digit buffers, it is different also with the time, depending on delaying as binary digit decoder, binary digit
The some parameters for the event rushed in device or digit buffer.However, it is not depend on by specific access and the bit quantity read
The actual bit read during particular access, this is the significant differences read with variable length codewords.Regular length position sequence
The reading of row is touched by some events in binary digit buffer, binary digit decoder or local digit buffer
Hair.As an example, when the bit quantity for being present in connected digit buffer is brought down below predetermined critical, can ask
New regular length bit sequence is read, wherein, different critical value can be used for not parity buffer.At encoder, it is necessary to ensure that Gu
Measured length bit sequence inserts bit stream with same sequence, wherein, they read in decoder-side from bit stream.Similar to foregoing
Embodiment, also can be by the intertexture of the fixed-length sequence and low latency control combination.In the following, it is described that it is used for fixed length
Spend the preferred embodiment of the intertexture of bit sequence.Obtained with reference to WO2011/128268A1 on the further of the latter's interleaving scheme
Details.
After being described above, description even previous coding is used for the embodiment of compressed video data, should
Embodiment is described as the even further embodiment for realizing embodiments of the invention, and the specific implementation causes
It is especially effective on the one hand compression ratio and the on the other hand good compromise between look-up table and computing overhead.Specifically, under
Row embodiment allows to carry out each bit stream of entropy code using the upper less complex variable-length code (VLC) of calculating, and effectively covers general
Rate estimates value part.In embodiment described below, symbol is binary properties, and the VLC codes presented as follows are effectively contained
Lid is by for example [0；0.5] R of extension inLPSThe probabilistic estimated value of expression.
Specifically, each entropy coder 310 and entropy decoding in the embodiment description difference Fig. 7 to Figure 17 being outlined below
Device 322.When they are appeared in image or video compression applications, they are applied to binary digit (that is, binary digit symbol)
Coding.Therefore, these embodiments are also applied for image or Video coding, wherein, such binary character is split off respectively
The bit stream 324 that the binary digit 307 being encoded is flowed and will be decoded into one or more, herein each such a binary system
Bit stream can be considered the realization of shellfish slave's profit processing routine (Bernoulli process).Embodiment described below is using hereafter
The multiple so-called variables explained to one or more of variable code (v2v- codes) carrys out coding binary bit stream.V2v- codes are visual
For with etc. quantity code word two unprefix codings.Primary and secondary unprefix coding.Each code of unprefix coding
Word is associated with a code word of secondary unprefix coding.According to the embodiment of following summary, encoder 310 and decoder 322
In at least some operations it is as follows：For the particular sequence of coding binary position 307, read once whenever from buffer 308
During the code word of unprefix coding, the corresponding code word of secondary unprefix coding is written to bit stream 312.Identical program is used to
Bit stream 324 as decoding, but primary and secondary unprefix coding exchanges.That is, for decoding bit stream 324, whenever from each
When bit stream 324 reads the code word of secondary unprefix coding, the corresponding code word of a unprefix coding is written to buffer
320。
Advantageously, code described below is without look-up table.The code is with finite state machine (finite state
Machine) form implements.Here, the v2v- codes can be produced by simple construction rule, thus, for code word,
Large-scale table need not be stored.But coding or decoding can be performed using simple algorithm.Three structure rules are described below,
Two of which can be parameterized.It covers the different or even disjunct parts at foregoing probability interval, and accordingly, thus
It is particularly advantageous when being used together, for example all three code uses (respectively for different coder/decoders 10 and 22) side by side,
Or both use wherein.Using aftermentioned structure rule, the conjunction of v2v- code collections can be designed so that for the shellfish with any probability P
Slave's profit processing, for long code length, an efficiency in these codes is good.
As described above, 312 and 324 each encoding and decoding of crossfire can be independent for each crossfire or entered with interleaving mode
OK.However, this is not specific for the v2v- codes classification, therefore, hereinafter, for each in three structure rules
The encoding and decoding of individual description particular code word.But it must emphasize that whole previous embodiments about intertexture solution also may be used
Combined with presently described code or coding and decoding device 310 and 322.
Structure rule 1：' unitary binary digit pipeline ' code or coder/decoder 310 and 322
Unitary binary digit pipeline code (PIPE=probability intervals partition entropy) is the spy of so-called ' binary digit pipeline ' code
Determine version (that is, the code suitable for encoding any of each bit stream 312 and 324), each transmission belongs to foregoing probability model
Enclose [0；The binary digit symbol statistical value data in some probability subinterval 0.5].The structure of binary digit pipeline code is described first
Build.Binary digit pipeline code can be from any unprefix coding structure with least three code words.In order to form v2v- codes, its
Using unprefix coding as a code and quadratic code, but two code words of secondary unprefix coding exchange.This represent except
Outside two code words, binary digit is written to unchanged bit stream.Using technique, an only unprefix coding must connect
Stored with the information, two code word exchanges, thus reduces the consumption of memory.Pay attention to, only exchange different length code word
To be reasonable, reason is, otherwise bit stream (will be ignored in bit stream end with the equal length with bit stream
The effect that may occur).
Due to structure rule, a prominent property of binary digit pipeline code is, if once with it is secondary unmatched
Sew code and exchange (while the mapping of code word retains), as a result gained v2v- codes and original v2v- code-phases are same.Therefore, enter for two
Pipeline code in position processed, encryption algorithm are identical with decoding algorithm.
Unitary binary digit pipeline code is built by specific unprefix coding.The specific unprefix coding is fabricated as follows.First,
The unprefix coding that is formed of n unary code word of generation, the n unary code word starts from ' 01 ', ' 001 ', ' 0001 ' ...
Until produce n code word.N is the parameter of unitary binary digit pipeline code.Tail end (trailing) is removed from most long code word
1.This, which corresponds to, blocks unitary code (but being free of code word ' 0 ').Then n-1 unary code word is generated, start from ' 10 ',
' 110 ', ' 1110 ' ... generation until produce n-1 code word.Tail end 0 is removed from most long code word.Two unprefix codings
Joint set be used as input to produce unitary binary digit pipeline code.Two code words exchanged are only formed by 0
One code word and only by 1 code word formed.
N=4 example：
Structure rule 2：' unitary to Lay this ' (' Unary to rice ') code and unitary are to this He of coder/decoder 10 of Lay
22：
Unitary uses Lay this yard and blocks unitary code as a code.That is, generation unary code word start from ' 1 ',
' 01 ', ' 001 ' ... until produce 2n+ 1 code word simultaneously removes tail end 1 from most long code word.N is ginseng of the unitary to this yard of Lay
Number.The code word of secondary unprefix coding from a unprefix coding is fabricated as follows.To only by 0 code word formed, dividing
With code word ' 1 '.All other code words two are entered by the n positions of the numeral 0 with the once respective code words without prefix code
The series connection of code word ' 0 ' shown of tabulating is formed.
N=3 example：
Pay attention to, it, which is equal to, maps to unlimited unitary code with this parameter (rice prameter) 2 of LaynLay this code
(rice code)。
Structure rule 3：' three binary digits ' (' Three bin ') code
Three binary digit code tables are shown as
It has following properties, and a code (symbol sebolic addressing) has regular length (being usually 3 binary digits), and
Code word is classified by the numeral 1 of rising.
Next effective specific implementation of three binary system bit codes of description.The encoder of three binary system bit codes and decoding
Device can be implemented without storage table in the following manner.
(any one 10) in the encoder, three binary digits are read from bit stream stream (that is, 7).If these three two enter
Position processed just includes one 1, then code word ' 1 ' is written to bit stream, is followed by the binary representation of the position by numeral 1
Two binary digits that (starting from right side 00) is formed.If three binary digits just contain one 0, code word ' 111 ' is write
To bit stream, two binary systems that the binary representation (starting from right side 00) of the position by numeral 0 is formed are followed by
Position.Remaining code word ' 000 ' and ' 111 ' is mapped to ' 0 ' and ' 11111 ' respectively.
In decoder (any one 22), bit (bin) or position (bit) is read from each bit stream 24.Such as
It is equal to ' 0 ' to fruit, then code word ' 000 ' is decoded into bit stream 21.If it is equal to ' 1 ', then from bit stream more than 24
Read two binary digits.If the two positions be not equal to ' 11 ', they be interpreted as numeral binary representation, two 0
And one 1 be decoded into bit stream so that 1 position is determined by numeral.If the two positions are equal to ' 11 ', there are two again
Position is read, and is interpreted as the binary representation of numeral.If the quantity is less than 3, two 1 and one 0 are decoded and count
Word determines 0 position.If equal to 3, then ' 111 ' it is decoded into bit stream.
Next effective specific implementation of description unitary binary digit pipeline code.Can be effectively by using counter
Realize the encoder and decoder for unitary binary digit pipeline code.Due to the structure of binary digit pipeline code, easily realize
The coding of binary digit pipeline code and decoding：
In encoder (any one 10), if the first binary digit of code word is equal to ' 0 ', binary digit is located
Manage until appearance ' 1 ', or until reading n 0 (first ' 0 ' containing the code word).If there is ' 1 ', then binary system is read
Position is written to unchanged bit stream.Otherwise (that is, read n 0), be then written to bit stream for n-1 1.If the code word
First binary digit is equal to ' 1 ', then binary digit is processed (contains the code word until occurring ' 0 ', or until reading n-1 1
First ' 1 ').If there is ' 0 ', then read binary digit and be written to unchanged bit stream.Otherwise (that is, n-1 is read
It is individual 1), then n 0 is written to bit stream.
In decoder (any one 322), identical algorithms are used for encoder, and reason is, this is for as described above two
System position pipeline code is identical.
Next effective specific implementation of the description unitary to this yard of Lay.Unitary can to the encoder and decoder of this yard of Lay
Effectively implemented, will now be described by using counter.
In encoder (any one 310), from bit stream (that is, 7) read binary digit until occur 1 or until
Read 2nUntill individual 0.Count 0 quantity.If count number is equal to 2n, then code word ' 1 ' be written to bit stream.Otherwise write
' 0 ', it is then the binary representation for the count number write as with n positions.
In decoder (any one 322), a position is read.If equal to ' 1 ', then 2nIndividual 0 is decoded into binary digit
Stream.If equal to ' 0 ', then there is more than n position to be read again and be interpreted as the binary representation of quantity.0 quantity is solved
Code is then ' 1 ' into bit stream.
In other words, the embodiment just described describes the encoder for coded identification sequence 303, including distributor
316, it is configured as the information that the inside of the prior symbol based on symbol sebolic addressing is included and multiple parameters 305 is distributed into the symbol
Each symbol of number sequence；Multiple entropy coders 310, each symbol for being configured as will be forwarded to each entropy coder 310
307 are converted into each bit stream 312；And selector 6, it is configured as each symbol 303 being forwarded to multiple entropy coders 310
In a selected entropy coder, the selection depends on distributing to the quantity of the parameter 305 of each symbol 303.According to firm
The embodiment of general introduction, at least the first entropy coder subset can be that (variable length coder is configured variable length coder
Respectively to map to and being inserted into the variable-length of bit stream 312 the variable-length symbol sebolic addressing in the crossfire of symbol 307
Code word, each entropy coder 310 of the first subset, according to this pair of mapping ruler, are had (2n-1) using bijective map rule
The code word of unprefix coding of >=3 code words is mapped to and the secondary unprefix coding code of a prefix code identical
Word, so that the whole than two of the code word of a unprefix coding is mapped to the identical of secondary unprefix coding
Code word, and two code words of primary and secondary unprefix coding have different length, and it is mapped onto one another in a manner of exchange,
Wherein different n can be used in entropy coder, so as to cover the different piece in the section of foregoing probability interval.First unprefix coding
It may be constructed in so that the code word of the first unprefix coding is (a, b)2、(a,a,b)3、…、(a,…,a,b)n、(a,…,a)n、
(b,a)2、(b,b,a)3、…、 (b,…,b,a)n-1、(b,…,b)n-1, and two code words mapped onto one another are in a manner of exchange
(a,…, a)nAnd (b ..., b)n-1, b ≠ a and a, b ∈ { 0,1 }.However, other selections are feasible.
In other words, each first subset of entropy coder can be configured as, and will be forwarded to the symbol of each entropy coder
When number being converted into each bit stream, check be forwarded to the first symbol of each entropy coder with determine (1) first symbol whether etc.
In a ∈ { 0,1 }, in this case, each entropy coder is configured to check for being transmitted to the follow-up symbol of each entropy coder
Number, with determine whether (1.1) follow occur b in the ensuing n-1 symbol after the first symbol and b ≠ a and b ∈
{ 0,1 }, in this case, each entropy coder be configured as by code word be written to each bit stream (its be equal to by forwarding
The first symbol followed to the successive character of each entropy coder), up-to symbol b；(1.2) follow the first symbol it
Do not occur b in ensuing n-1 symbol afterwards, in this case, each entropy coder is configured as code word being written to
Each bit stream, it is equal to (b ..., b)n-1；Or (2) first symbol be equal to b, in this case, each entropy coder by with
The successive character for checking and being forwarded to each entropy coder is set to, and determines whether (2.1) and is following connecing after the first symbol
There is a in the n-2 symbol to get off, in this case, each entropy coder is configured as code word being written to each bit
Stream, it is equal to the successive character by being forwarded to each entropy coder until the first symbol that symbol a is followed；Or (2.2) with
With not occurring a in the ensuing n-2 symbol after the first symbol, in this case, each entropy coder is configured
For code word is written into each bit stream, its be equal to (a ..., a)n。
Additionally or alternatively, the yield in the second subset of entropy coder 10 can be variable length coder, and it is configured as can
Become length symbol sebolic addressing and respectively map to fixed length codeword, the entropy coder of yield in the second subset each uses bijective map
Rule, according to this pair of mapping ruler, have and belong to { (a), (ba), (bba) ..., (b ... ba), (bb ... b) } and b ≠ a and a, b
The 2 of ∈ { 0,1 } typenThe code word for once blocking unitary code of+1 code word is mapped to the code word of secondary unprefix coding,
So that the code word (bb ... b) for once blocking unitary code is mapped to the code word (c) of secondary unprefix coding；And once block
The code word of unitary code the other code words of whole (a), (ba), (bba) ..., (b ... ba) be mapped to (d) and c
≠ d and c, d ∈ { 0,1 } are as the code word of prefix and n positions word as suffix, wherein, entropy coder uses different n.Entropy is compiled
The yield in the second subset of code device each can be configured such that n positions word is the b quantity in each code word for once block unitary code
N positions represent.But other selections are also feasible.
Again, can be each configured as from the viewpoint of the operator scheme of each encoder 10, the yield in the second subset of entropy coder
The sign reversing of each entropy coder will be forwarded into each bit stream, counting and be forwarded to the symbol of each entropy coder
B quantity in number sequence, until a occurs, or until the quantity of the symbol sebolic addressing of each entropy coder is forwarded to up to 2n, and be somebody's turn to do
Whole the 2 of sequencenSymbol is b, and (1) if b quantity is equal to 2n, then by c and c ∈ { 0,1 } are as secondary unprefix coding
Code word is written to each bit stream；And (2) if b quantity be less than 2n, then the code word of secondary unprefix coding is written to respectively
Individual bit stream, it has (d) and c ≠ d and d ∈ { 0,1 } be used as prefix, and according to b quantity determination n positions word as suffix.
Equally, additionally or alternatively, the predetermined entropy coder in entropy coder 10 can be variable length code
Device, it is configured as regular length symbol sebolic addressing respectively mapping to variable length codewords, and the predetermined entropy coder uses
Bijective map rule, according to the 2 of the length 3 of the regular code of the bijective map3Code word is mapped to secondary unprefix coding
Code word so that the code word (aaa) of a code3(a ∈ { 0,1 }) is mapped to code word (c) (c ∈ { 0,1 }), tool
All three code words for having proper what a b (b ≠ a and b ∈ { 0,1 }) code be mapped to (d) (c ≠ d and
D ∈ { 0,1 }) code word of each first 2 word as suffix outside prefix, and the first set with 2- words is used as,
All three code words of a proper code with an a are mapped to is used as prefix with (d), and has and do not belong to first
One 2 word of set element and the code word connected as suffix of or not 2nd 2 word in the 2nd 2 word set；And
Wherein code word (bbb)3It is mapped to and is used as prefix with (d), and with the one 2 word for being not belonging to first set element
With the code word connected as suffix for the 2nd 2 word for being not belonging to second set element.Just there is a b code
Code word the one 2 word can be b positions in each code word of a code 2 expressions；And just there is one
2nd 2 word of the code word of an a code can be 2 expressions of a positions in each code word of a code.
But other selections are also feasible.
Again, the predetermined entropy coder in entropy coder can be configured as that predetermined entropy coder will be forwarded to
Symbol be converted into each bit stream, checked in a manner of 3 one group deliver to predetermined entropy coder symbol whether (1) 3
Symbol is made up of a, and in this case, predetermined entropy coder is configured as code word (c) being written to each bit stream；(2)3
Individual symbol contains a b, in this case, predetermined entropy coder be configured as using with (d) as prefix and with three
2 code words for being denoted as suffix of b positions are written to each bit stream in character；(3) three characters just include an a,
In this case, it is the 1st of first set element as prefix and not that predetermined entropy coder, which is configured as with (d),
Position word is written to each bit stream with 2 connecting for expression of a positions in 3 characters as the code word of suffix；Or (4) by b groups
Into 33 characters, in this case, it as prefix and is not the first collection that predetermined entropy coder, which is configured as with (d),
One 2 word of the element of conjunction is written to being not belonging to the connecting for the 2nd 2 word of second set element as the code word of suffix
Each bit stream.
On decoding side, the embodiment just described discloses a kind of decoder for being used for rebuilding the sequence of symbol 326, bag
Containing multiple entropy decoders 322, it is each configured as each bit stream 324 being converted into symbol 321；Distributor 316, its quilt
The symbol information contained for the symbol sebolic addressing first rebuild before being configured to, multiple parameters are distributed to reconstructed symbol sequence
Each symbol 326 of row；And selector 318, it is configured as recovering from the selected entropy decoder in multiple entropy decoders
By each symbol 325 of reconstructed symbol sebolic addressing, the selection depends on being defined to the number of parameters of each symbol.According to firm
The embodiment just described, at least the first subset of entropy decoder 322 is variable-length decoder, and it is configured as variable length
Degree code word respectively maps to variable-length symbol sebolic addressing, and each entropy decoder 22 of the first subset uses bijective map
Rule, according to bijective map rule, the code word with a unprefix coding of (2n-1) >=3 code word is mapped to
With the secondary unprefix coding code word of a prefix code identical so that the code word of a unprefix coding except two it
Outer whole is mapped to the same code word of secondary unprefix coding, and two code words of primary and secondary unprefix coding have
There is different length, and it is mapped onto one another in a manner of exchange, and wherein different n can be used in entropy coder.First unprefix coding can quilt
The code word for being constructed so that the first unprefix coding is (a, b)2、(a,a, b)3、…、(a,…,a,b)n、(a,…,a)n、(b,
a)2、(b,b,a)3、…、(b,…,b,a)n-1、(b,…, b)n-1, and two code words mapped onto one another are in a manner of exchange
(a,…,a)nAnd (b ..., b)n-1, b ≠ a and a, b ∈ { 0,1 }.However, other selections are feasible.
First subset of entropy coder can be each configured as, and be converted into by each bit stream in symbol, be checked each
First of bit stream determines whether：(1) first is equal to a ∈ { 0,1 }, in this case, each entropy coder by with
The subsequent bit for checking and being transmitted to each bit stream is set to, is determined whether：(1.1) the following n-1 position after first
There is b (and b ≠ a and a, b ∈ { 0,1 }), in this case, each entropy decoder is configured as rebuilding symbol sebolic addressing, its etc.
In first, the subsequent bit followed by for each bit stream is until position b；Or the ensuing n-1 of (1.2) after first
Individual position does not occur b, and in this case, each entropy decoder is configured as rebuilding symbol sebolic addressing, and it is equal to (b ..., b)n-1；
Or (2) first be equal to b, in this case, each entropy decoder is configured to check for the subsequent bit of each bit stream, and
Determine whether：(2.1) there is a in the ensuing n-2 position after first, in this case, each entropy decoder by with
Reconstruction symbol sebolic addressing is set to, it is equal to first, and followed by the subsequent bit for each bit stream is until position a；Or (2.2) exist
Ensuing n-2 position after first does not occur a, and in this case, each entropy decoder is configured as rebuilding symbol sequence
Row, its be equal to (a ..., a)n。
Additionally or alternatively, at least the yield in the second subset of entropy decoder 322 can be variable-length decoder, and it is configured
For fixed length codeword respectively to be mapped to variable length character sequence, the entropy decoder of yield in the second subset is each using double
Mapping ruler is penetrated, according to bijective map rule, the code word of the second unprefix coding is mapped to the generation for once blocking unitary code
Code word (have and belong to (a), (ba), (bba) ..., (b ... ba), (bb ... b) } and b ≠ a and a, the 2 of b ∈ { 0,1 } typen+1
Individual code word) so that the code word (c) of secondary unprefix coding is mapped to the code word (bb ... b) for once blocking unitary code；
And there is (d) and c ≠ d and c, d ∈ { 0,1 } are mapped to as the code word of suffix as prefix and n positions word and once block one
Other code words { the respective code word in { (a), (ba), (bba) ..., (b ... ba) }, wherein entropy of the code word of first code
Decoder uses different n.Each of the yield in the second subset of entropy decoder can be configured such that n positions word once blocks to be each
The n positions of b quantity in the code word of unitary code represent.But other selections are also feasible.
Each in each yield in the second subset of entropy decoder can be variable-length decoder, and it is configured as will be solid
Measured length code word respectively maps to variable-length symbol sebolic addressing, and its is configured as by the bit of each entropy decoder
Circulation is changed into symbol, is checked first of each bit stream and is determined whether：(1) it is equal to c and c ∈ { 0,1 }, this
In the case of, each entropy decoder, which is configured as rebuilding, is equal to (bb ... b)2N symbol sebolic addressing and b ∈ { 0,1 }；Or (2) its be equal to
D and c ≠ d and c, d ∈ { 0,1 }, in this case, each entropy decoder are configured as from each bit stream at first
Position determines n positions word to n of rear in addition, and rebuilds symbol sebolic addressing therefrom, and it belongs to { (a), (ba), (bba) ..., (b ...
Ba), (bb ... b) } type and b ≠ a and b ∈ { 0,1 }, b quantity depend on n positions word.
Additionally or alternatively, the predetermined entropy decoder in entropy decoder 322 can be variable-length decoder,
It is configured as respectively mapping to variable length codewords into regular length symbol sebolic addressing, and the predetermined entropy decoder uses double
Mapping ruler is penetrated, according to bijective map rule, the code word of secondary unprefix coding is mapped to the length 3 of a code
23Individual code word so that code word (c) and c ∈ { 0,1 } are mapped to the code word (aaa) of a code3And a ∈ { 0,1 }；Tool
There is (d) and c ≠ d and d ∈ { 0,1 } is as each first 2 word conduct outside prefix and the first set of three 2 words
The code word of suffix is mapped to all three code words of a code with a proper b, and b ≠ a and b ∈ { 0,1 }；
With (d) as prefix and with the one 2 word for being not belonging to first set element and the second set of 32 words outside
The series connection of 2nd 2 word is mapped to all three codes of a code just with an a as the code word of suffix
Word；And and as prefix and with the one 2 word for being not belonging to first set element be not belonging in second set with (d)
The series connection of the 2nd 2 word of element be mapped to code word (bbb) as the code word of suffix3.Just there is a b once
One 2 word of the code word of code can be 2 expressions of the b positions in each code word；And just there is one
2nd 2 word of the code word of an a code can be 2 expressions of a positions in each code word.But
Other selections are also feasible.
The entropy decoder that makes a reservation for of entropy decoder can be a kind of variable-length decoder, and it is configured as will be variable
Length code word respectively maps to each symbol sebolic addressing for having three symbols by oneself；And its it is configured as each entropy decoder
Bit stream is converted into symbol, checks first of each bit stream to determine whether：(1) first etc. of each bit stream
In c and c ∈ { 0,1 }, in this case, predetermined entropy decoder is configured as rebuilding symbol sebolic addressing, and it is equal to (aaa)3And a
∈{0,1}；Or first of (2) each bit stream is equal to d and c ≠ d and d ∈ { 0,1 }, in this case, predetermined entropy solution
Code device, which is configured as being connected on two other of first rear from each bit stream, determines the one 2 word, and check this first
2- positions word determines whether：(2.1) the first 2- positions words are not belonging to the element in the first set of three 2 words, this
In the case of, the predetermined entropy decoder is configured as rebuilding symbol sebolic addressing, its b ≠ a and b ∈ { 0,1 }, b with what a proper b
The one 2 word is depended in the position of each symbol sebolic addressing；Or (2.2) the 1st words are the element in first set,
In this case, the predetermined entropy decoder be configured as two positions (the one 2 word is had determined from the two positions) it
Afterwards, the 2nd 2- positions word is determined from another two of each bit stream, and checks the 2nd 2 word to determine whether (3.1) second
2 words are not belonging to the element of the second set in three 2 words, and in this case, predetermined entropy decoder is configured as rebuilding
Just there is a symbol sebolic addressing, and a depends on the 2nd 2 word in the position of each symbol sebolic addressing；Or (3.2) the 2nd 2 words
It is the element of the second set in three 2 words, in this case, predetermined entropy decoder is configured as reconstruction and is equal to
(bbb)3Symbol sebolic addressing.
Now, after the universal of description Video Coding Scheme, relative to above embodiment described the present invention's
Embodiment.In other words, the embodiment being summarized below can realize that vice versa using such scheme, above-mentioned encoding scheme
It can use and be realized using the embodiment being summarized below.
In the above-described embodiment described for Fig. 7-9, Fig. 1-6 entropy coder and decoder according to PIPE concepts come
Realize.One specific embodiments uses arithmetic list Probability State coder/decoder 310 and 322.As described below, according to optional reality
Example is applied, entity 306-310 and correspondent entity 318-322 can be replaced with general entropy code engine.As example, the imagination is calculated
Art encodes engine, and it manages a shared state R with L and by all symbolic codings into a shared bit stream, thus abandons
On parallel processing existing PIPE concepts it is favourable in terms of, but avoid the necessity of interlaced sections bit stream, it is such as following
Discussed further.So, the quantity of the Probability State utilized by updating the probability of (for example tabling look-up) estimation context can
With more than the quantity for carrying out the Probability State that probability interval subdivision is utilized.That is, similar to index table Rtab it
Preceding quantization probability interval width value, it can equally quantify Probability State index.Above to single coder/decoder 310 and 322
Description in the cards therefore can be expanded, using as entropy coding/decoding device 318-322/306-310 as context oneself
The example of the binary arithmetic coding/decoding engine implementation of adaptation.
More precisely, according to embodiment, the defeated of parametric distribution device (it serves as context distributor herein) is attached to
The entropy coder gone out can be operated in the following manner：
0th, distributor 304 forwards binary bit value together with probability parameter.Probability is pState_current [bin].
1st, therefore, entropy code engine receives：1) valLPS, 2) binary digit and 3) Distribution estimation value pState_
current[bin].PState_current [bin] can have the more of the distinguishable Probability State index than Rtab
State.If it were to be so, pState_current [bin] for example (can be more than or equal to 1 by ignoring m LSB and m
And preferably 2 or 3) quantified to obtain p_state, that is, to be subsequently used for access table Rtab index.However, it is possible to
Without the quantization, i.e. p_state can be pState_current [bin].
2 then R is quantified (as described above：One of R (and corresponding L with a shared bit stream)
All distinguishable values/for p_state all distinguishable values for p_state are managed, or per p_state
Distinguishable one R of value (and corresponding L with associated section bit stream relative to every R/L), the latter's such case pair
It should be worth in bit encoder 310 per such)
Q_index=Qtab [R>>Q] (or some other quantized versions)
3rd, it is then determined that RLPSAnd R：
RLPS=Rtab [p_state] [q_index]；Rtab stores p [p_state] Q [q_index] wherein
Pre-calculated values
R=R-RLPS【That is, R is tentatively updated in advance, just as " binary digit " is MPS】
4th, the calculating in new portion section：
If (bin=1-valMPS),
5th, L and R standardization again, position is write,
Similarly, the entropy decoder for being attached to parametric distribution device (it serves as context distributor herein) can be following
Mode is operated：
0th, distributor 304 forwards binary bit value together with probability parameter.Probability is pState_current [bin].
1st, therefore, entropy decoding engine receives the request of binary digit together with the following：1) valLPS and 2) probability
It is distributed estimate pState_current [bin].PState_current [bin] can have the distinguishable probability than Rtab
The more state of state index.If it were to be so, pState_current [bin] for example can be by ignoring m
Individual LSB (m is more than or equal to 1 and preferably 2 or 3) is quantified to obtain p_state, that is, to be subsequently used for access table
Rtab index.However, it is possible to without the quantization, i.e. p_state can be pState_current [bin].
2 then R is quantified (as described above：One of R (and it is corresponding with shared bit stream
V) all distinguishable values/for p_state all distinguishable values for p_state are managed, or per p_state
One R of distinguishable value (and relative to every R/L couples of the corresponding V with associated section bit stream), this feelings of the latter
Condition corresponds to, and there is bit encoder 310 to be worth per such)
Q_index=Qtab [R>>Q] (or some other quantized versions)
3rd, it is then determined that RLPSAnd R：
RLPS=Rtab [p_state] [q_index]；Rtab stores p [p_state] Q [q_index] wherein
Pre-calculated values
R=R-RLPS【That is, R is tentatively updated in advance, just as " binary digit " is MPS】
4th, binary digit is determined according to the position of partial section：
if(V3R), then
(binary digit is decoded as LPS；Binary digit buffer selector 18 will pass through
Actual binary place value is obtained using this binary piece of information and valMPS)
else
(binary digit is decoded as MPS；By using this binary piece of information and valMPS
Obtain actual binary place value)
5th, R standardization again, read a position and update V,
As described above, pState_current [bin] is distributed to each binary digit by distributor 4.Correlation is based on
Context selection is carried out.That is, distributor 4 can select context using context index ctxIdx, the context
Thus index has respective pState_current related to this.Whenever it can carry out probability updating, probability
PState_current [bin] is applied to current binary digit.Probability State pState_current [bin] is more
New root is carried out according to the value of bits of coded：
If (bit=1-valMPS), then
pState_current←Next_State_LPS[pState_current]
- the valMPS of if (pState_current=0), then valMPS ← 1
else
pState_current←Next_State_MPS[pState_current]
Provided that more than one context, then adjusted in a manner of context, i.e. pState_current
[ctxIdx] is used to encode, and is then updated using current binary bit value and (is separately encoded or decoded).
General introduction as discussed, according to the embodiment that will now be described, encoder and decoder can optionally be embodied as
Operate in different modes, i.e. low-complexity (LC) pattern and high efficiency (HE) pattern.Hereinafter compiled mainly for PIPE
Code shows (and then referring to LC and HE PIPE patterns), but the description of the details of complexity changeable is easy to be transferred to entropy
Other realizations of coding/decoding engine are such as using the embodiment of a shared context-adaptive arithmetic coder/decoder.
According to the embodiment being summarized below, the two entropy code patterns can be shared：
Identical syntax and semantics (respectively for syntactic element sequence 301 and 327)
Identical binarization scheme (as currently specified for CABAC) (that is, binarization of all syntactic elements
Device can be operated, unrelated with the pattern of activation)
Identical PIPE codes use (that is, binary digit coder/decoder can be operated, with the pattern of activation without
Close)
The use (rather than current 16 initialization values specified for CABAC) of 8 probabilistic model initialization values
In general, LC-PIPE and HE-PIPE difference is to handle complexity, for example is each binary digit
Select the complexity in PIPE paths 312.
For example, LC patterns can operate under following constraint：For each binary digit (binIdx), it is understood that there may be just
What a probabilistic model a, i.e. ctxIdx.That is, it is not possible to which context selection/adjustment is provided in LC PIPE.
The syntactic element that specific syntactic element is such as used for remaining coding can be encoded using context, such as further below
General introduction.Moreover, all probabilistic models can be non-self-adapting, i.e. all models can be with the every of appropriate model probability
Individual fragment is initialised (according to clip types and fragment QP selection) when starting and can be during the whole processing of fragment
Keep fixing.For example, for both context modeling and coding, the correspondences of PIPE codes 310/322 different from 8 may only be supported
8 different model probabilities.The concrete syntax element (its semantic general introduction in greater detail below) encoded for remnants, i.e.,
Significance_coeff_flag and coeff_abs_level_greaterX (wherein X=1,2), can distribute to probability
Model, so that (at least) group of such as 4 syntactic elements is encoded/decoded using identical model probability.With
CAVLC is compared, and LC-PIPE patterns reach roughly the same R-D performances and identical handling capacity.
HE-PIPE is configurable to be similar to CABAC H.264 in concept, has following difference：Two enter
Arithmetic coding (BAC) processed is encoded with PIPE and replaces (identical with the case of LC-PIPE).Each probabilistic model, i.e., it is each
CtxIdx, it can be represented with pipeIdx and refineIdx, wherein the pipeIdx with the value in the range of 0-7 represents 8 not
With the model probability of PIPE codes.The change can only influence internal state expression, behavior (that is, the probability without influenceing state machine
Estimation) itself.General introduction as discussed, the initialization of probabilistic model can use 8 initialization values as described above.
Syntactic element coeff_abs_level_greaterX (wherein X=1,2), coeff_abs_level_minus3 and coeff_
Sign_flag (its semanteme will be apparent from according to following discussion) scanning backward can be swept along with forward scan identical
Retouch path and carry out (for example, being used for significant-image coding).It can also simplify for encoding coeff_abs_level_greaterX
The context of (wherein X=1,2) derives.Compared with CABAC, the HE-PIPE proposed is reached substantially with more preferable handling capacity
Identical R-D performances.
It is readily apparent that the pattern just referred to is for example by providing aforementioned context adaptive binary arithmetic volume/solution
Code engine generates easily so that the engine operates in different modes.
Therefore, according to the embodiment for meeting the first aspect of the present invention, the decoder for decoding data stream can construct
Into as shown in figure 11.Decoder is used for decoding data stream 401, such as interleaved bitstream 340 (media data, such as video data
It is encoded into the data flow).Decoder includes mode switch 400, and it is configured to activate low complexity according to data flow 401
Sexual norm or high efficiency mode.Therefore, data flow 401 can include syntactic element such as binary syntax element, in low complexity
When pattern is the pattern to be activated, it is 1 to have binary value, when high efficiency mode is the pattern to be activated, has two to enter
Value processed is 0.Obviously, the correlation between binary value and coding mode can be switched, can also use has two or more
The nonbinary syntactic element of possible values.Because actual selection between two patterns before each syntactic element is received still
It is unclear, so the syntactic element may be embodied in the data flow for example encoded using fixation probability estimate or probabilistic model
In 401 some forward position header, or data flow 401 is write as carried out using bypass mode.
Further, Figure 11 decoder includes multiple entropy decoders 322, and each of which is configured to data flow 401
In code word be converted into the partial sequence 321 of symbol.As described above, the one side of deinterleaver 404 can be connected to entropy decoding
Between the input of device 322, between the input for the decoder that on the other hand can be connected to the applicable Figure 11 of data flow 401.Enter one
Step ground, as already described above, each of entropy decoder 322 can be associated with each probability interval, in entropy decoder
In the case of 322 processing MPS and LPS rather than absolute symbol value, the probability interval of each entropy decoder cover together 0-1 or
0-0.5 whole probability interval.Details on this problem have been carried out describing above.Then, it is assumed that decoder 322
Quantity be 8, and PIPE indexes are assigned to each decoder, but arbitrarily other quantity is also feasible.Further, this
One in a little encoders, (it is exemplarily one with pipe_id 0 that hereinafter, this, which is) is to be directed to have to wait generally
The binary digit of rate statistics optimizes, i.e. their binary bit value is equally possible to be assumed to be 1 and 0.The decoder is only
To be transmitted in binary digit.The operation of each encoder 310 is identical.Maximum probability is even depended on by selector 402 and 502
The operation of any binary digit that binary bit value (valMPS) is carried out respectively, can be without.In other words, each several part crossfire
Entropy be optimization.
Further, Figure 11 decoder includes selector 402, is configured to from multiple entropy decoders 322 selected one
Each symbol of individual retrieval symbol sebolic addressing 326.As described above, selector 402 can be divided into parametric distribution device 316 and selection
Device 318.Inverse symbolism device 314 is configured to inverse symbolism symbol sebolic addressing 326 to obtain syntactic element sequence 327.Reconstructor
404, which are configured to syntactic element sequence 327, carrys out re-constructing media data 405.Selector 402 is configured to according to such as the institute of arrow 406
Activation pattern in the low-complexity pattern and high efficiency mode of instruction is selected.
As has already been mentioned above, reconstructor 404 can be fixation syntax and semantics to syntactic element (i.e., relatively
Fixed in the model selection that mode switch 400 is carried out) part of the Video Decoder based on prediction block that works.
That is, the construction of reconstructor 404 will not be influenceed by mode switch ability.More precisely, reconstructor 404 due to
The mode switch ability and realize expense without increasing at least feature of residual data that mode switch 400 provides, and
Regardless of the pattern for switching 400 selections, prediction data all keeps identical.However, this is same suitable for entropy decoder 322
With.All these decoders 322 are all reused under the two patterns, although correspondingly Figure 11 decoder with this two
Individual pattern (low-complexity pattern and high efficiency mode) is all compatible, but also realizes expense in the absence of extra.
On the other hand, it is noted that Figure 11 decoder is not only able under a pattern or another pattern to certainly
Worked comprising data flow.On the contrary, Figure 11 decoder and data flow 401 may be arranged so that between the two patterns
It is even and possible such as during video or some audio-frequency unit during switching in a media data, so as to such as root
The codec complexity of decoding side is controlled according to outside or environmental condition such as battery status etc., wherein utilizing from decoder to volume
The feedback channel of code device selects so as to corresponding lock ring control model.
Therefore, Figure 11 decoder in both cases, i.e., in the case where selecting LC patterns or is selecting HE patterns
In the case of operation it is similar.Reconstructor 404 is rebuild using syntactic element and by handling or obeying some syntactic structures
Regulation ask the current syntax element of predetermined syntax element type.Inverse symbolism device 314 ask several binary digits so as to
Produce the effective binarization for the syntactic element that reconstructor 404 is asked.Obviously, for binary alphabet, by inverse symbol
The binarization that number change device 314 is carried out is reduced to only passes to reconstructor 404 as currently using each binary digit/symbol 326
The binary syntax element asked.
However, the independent action of selector 402 is in the pattern selected by mode switch 400.The operator scheme of selector 402
Often more complicated for high efficiency mode, complexity reduces for low-complexity pattern.Moreover, following discussion table
It is bright during continuous symbol is retrieved from entropy decoder 322, operator scheme of the selector 402 under less complicated pattern
It often can equally reduce the speed that selector 402 changes the selection between entropy decoder 322.In other words, in low-complexity mould
Under formula, the probability increase of the close continuous symbol retrieved from the identical entropy decoder in multiple entropy decoders 322.This by
This allows quickly to retrieve symbol from entropy decoder 322.Under high efficiency mode, the operator scheme of selector 402 thereby tends to
It can cause to be selected between entropy decoder 322, wherein the probability being more closely associated with each selected entropy decoder 322
Section is adapted to the actual symbol statistics for the symbol currently retrieved by selector 402, is thus generating each number according to high efficiency mode
According to during stream more preferable compression ratio is produced in coding side.
For example, different behaviors of the selector 402 under the two patterns can realize it is as follows.For example, selector 402 can
To be configured to be directed to predetermined symbol, in the case where high efficiency mode is activated according to the prior search symbol of symbol sebolic addressing 326
And in the case that low-complexity is activated not according to the prior search symbol of symbol sebolic addressing between multiple entropy decoders 322
Selected.Dependence to the prior search symbol of symbol sebolic addressing 326 can be adaptive by context-adaptive and/or probability
Answering property causes.Both adaptivitys can close during low-complexity pattern in selector 402.
According to further embodiment, data flow 401 can be configured to continuous part such as fragment, frame, group of picture, frame
Sequence etc., and each symbol of symbol sebolic addressing can be each associated with multiple sign patterns.In this case,
Selector 402 is configurable to the feelings being activated for the symbol of the predetermined symbol type in current portions in high efficiency mode
Selection is changed according to the prior search symbol of the symbol sebolic addressing of the predetermined symbol type in current portions under condition, and low multiple
Polygamy pattern keeps selection invariable in current portions in the case of being activated.That is, selector can be allowed
402 change predetermined symbol types entropy decoder 322 between selection, but these change be confined to occur continuous part it
Between conversion between.Using this measure, the assessment of actual symbol statistics is confined to infrequent time instance, and simultaneously
Codec complexity is reduced in most of time.
Further, each symbol of symbol sebolic addressing 326 can be each associated with multiple sign patterns, and selects
Device 402 is selected to be configurable to be selected according to the prior search symbol of symbol sebolic addressing 326 for the predetermined symbol of predetermined symbol type
One in multiple contexts is selected, basis and selected context-sensitive probability in the case where high efficiency mode is activated
Model and according to predetermined symbol renewal and selected context-sensitive probabilistic model selected between entropy decoder 322
Select, one in multiple contexts is selected according to the prior search symbol of symbol sebolic addressing 326, and in low-complexity pattern quilt
Basis and selected context-sensitive probabilistic model and holding and selected context-sensitive probability in the case of activation
Model is constant to be selected between entropy decoder 322.That is, selector 402 can be directed to certain under both patterns
One syntax element type and use context-adaptive, while suppress probability adaptation in the case of LC patterns.
Alternatively, selector 402 may only reduce the renewal rate of the probability adaptation of LC patterns relative to HE patterns,
Without completely inhibiting probability adaptation.
Further, in other words, it is possible to LC pipelines particular aspects (i.e. the aspect of LC patterns) be described as follows.Specifically
Ground, non-self-adapting probabilistic model can use under LC patterns.Non-self-adapting probabilistic model can have hard coded (i.e. overall
Invariable probability or its probability only keep fixing during the entire process of fragment is handled), and therefore can be according to fragment
Type and QP (that is, such as the quantization parameter of signal transmission in the data flow 401 of each fragment) are configured.By assuming that
Distribute to the continuous binary digit sign-following fixation probability model of same context, can in one step to these two
Some in system position are decoded, because such binary digit utilizes identical pipeline code, that is, utilize identical entropy decoder
Encoded, and omit the probability updating after each decoding binary digit.Omit probability updating and save coding and decoding
During operation, therefore also result in complexity reduce and hardware design be largely simplified.
Non-self-adapting constraint can be wiped for all or part of selected probabilistic model so that a certain amount of two
System position allows probability updating after being encoded/decoded using the model.Suitable renewal section allows probability adaptive
And it can should at once decode some binary digits simultaneously.
Hereinafter, the more detailed of possible shared and complexity changeable the aspect of LC pipelines and HE pipelines is presented
Description.Specifically, hereinafter, various aspects are described, these aspects can in an identical manner or can with complexity
The mode of increase and decrease is used for LC pipelines pattern and HE pipeline patterns.Complexity changeable refers to that LC situations are special by removing
Fixed part is replaced them and drawn from HE situations by using less complicated thing.However, continuing it
Before, it should be mentioned that, Figure 11 embodiment is easy to be transferred to above-mentioned context adaptive binary arithmetic coding/decoding implementation
Example：Selector 402 and entropy decoder 322, which are compressed into, can directly receive data flow 401 and select what is currently drawn from data flow
The context adaptive binary arithmetic decoding device of the context of binary digit.This is adaptive to context-adaptive and/or probability
It is especially real for answering.During low-complexity pattern, function/adaptivity can close, or design is lighter.
For example, during Figure 11 embodiment is realized, being related to the pipeline entropy code stage of entropy decoder 322 can make
Variable coded system (sytematic variable-to-variable-codes) can be changed to eight kinds of systematicness, i.e. every
Individual entropy decoder 322 may belong to the v2v types having been described above.PIPE Coded concepts using systemic v2v- codes lead to
The quantity for crossing limitation v2v- codes simplifies.In the case of context adaptive binary arithmetic decoding device, this can equally be managed
The equal probabilities state of different contexts simultaneously segments itself or its quantised versions for probability.By CABAC or probabilistic model shape
State (state i.e. for probability updating) is mapped to PIPE ID or probability index to be searched in Rtab as described in Table A.
Table A：CABAC states to PIPE indexes mapping
The amended encoding scheme may be used as the foundation of the method for video coding of complexity changeable.It is general when carrying out
During rate mode adaptive, selector 402 or context adaptive binary arithmetic decoding device select PIPE decoders 322 respectively,
I.e. based on the Probability State index (be here illustratively 0-62) associated with the symbol currently to be decoded such as via upper and lower
The pipeline index and probability index that text will be used respectively using the mapping shown in Table A import Rtab, can be according to current decoding
The specific table row that is for example indexed using the next Probability State that to be accessed respectively in the case of MPS and LPS is pointed to of symbol
Conversion value is walked to update Probability State index., can be without renewal later in the case of LC patterns.Even in the overall situation
Can be without mapping in the case of fixed probabilistic model.
However, can be used arbitrary entropy code set and the technology in the document can also with it is secondary it is adaptive together with make
With.
Figure 11 above description is conversely commonly referred to as syntactic element and syntax element type.Hereinafter, will be to conversion
The complexity configurable code of coefficient level is described.
For example, reconstructor 404 is configurable to based on independently of the high efficiency mode or low-complexity pattern to be activated
A part for syntactic element sequence rebuilds the transform blockiis 200 of transformation series several levels 202, this part of syntactic element sequence 327
Include the important of the significance map of the position of the non-zero transform coefficient several levels in restriction instruction transform blockiis 200 in a manner of without intertexture
The level syntactic element of graph grammar element, then (followed by) restriction non-zero transform coefficient several levels.Specifically, following member can be related to
Element：Indicate the end position syntactic element (last_ of the position of last non-zero transform coefficient several levels in transform blockiis
significant_pos_x,last_significant_pos_y)；Significance map is limited together and for along in transform blockiis
(200) each position of each position instruction in the one-dimensional path (274) of last non-zero transform coefficient several levels is led in from DC positions
The transformation series several levels put whether the first syntactic element (coeff_significant_flag) of non-zero；For one-dimensional path
(274) each position (non-zero transform coefficient several levels wherein are located according to the first binary syntax element) indicates each position
Transformation series several levels whether be more than one the second syntactic element (coeff_abs_greater1)；For each of one-dimensional path
Position (the transformation series several levels more than one are wherein located according to the first binary syntax element) shows each of each position
Transformation series several levels the 3rd syntactic element (coeff_abs_greater2, coeff_abs_minus3) how many more than 1.
Order between end position syntactic element, first, second, and third syntactic element is to high efficiency mode and low complexity
Can be identical for sexual norm, and selector 402 is configurable to therefrom obtaining end for inverse symbolism device 314
Positional grmmar element, the first syntactic element, the second syntactic element and/or the 3rd syntactic element symbol entropy decoder 322 it
Between selected, high efficiency mode is still activated and different according to activation low-complexity pattern.
Specifically, selector 402 is configurable to, and the first syntactic element and are therefrom obtained for inverse symbolism device 314
The symbol of predetermined symbol type in the symbol subsequence of two syntactic elements, the predetermined symbol type in symbol subsequence
Prior search symbol select one in multiple contexts and efficient in activation to be directed to each symbol of predetermined symbol type
Basis is selected with selected context-sensitive probabilistic model in the case of rate pattern, in activation low-complexity pattern
In the case of by be segmented it is constant in a manner of be selected so that selection is kept not in the subdivision of continuously subsequence
Become.As described above, subdivision can measure in multiple positions, when being measured along one-dimensional path 274, each subdivision is described
Extend on position, or measured in using all types of multiple syntactic elements of current context coding.Namely
Say, binary syntax element coeff_significant_flag, coeff_abs_greater1 and coeff_abs_
Greater2 etc. is adaptive coding context, wherein the probabilistic model based on context selected under HE patterns selects decoding
Device 322.Also use probability adaptation.Under LC patterns, also enter with different contexts, the context for each two
Syntactic element coeff_significant_flag, coeff_abs_greater1 and coeff_abs_greater2 processed.So
And for each of these syntactic elements, context is directed to and keeps static along the Part I in path 274, only in conversion under
At one, and then change context along the part in path 274.For example, each part can be defined to block 200 4,8,
16 position length, whether there is unrelated in each position with each syntactic element.For example, coeff_abs_greater1 and coeff_
Abs_greater2 exists only for critical positions (that is, coeff_significant_flag is 1 position).Alternatively, often
Whether individual part can be defined to 4,8,16 syntactic element length, with resulting each several part on more block locations
Extend unrelated.For example, coeff_abs_greater1 and coeff_abs_greater2 exists only for critical positions, therefore,
Four syntactic elements partially due to can each prolong along the position therebetween in path 274 on more than 4 block locations
Stretch, not transmitting this syntactic element for it does not have coeff_abs_greater1 and coeff_abs_greater2 such as,
Because the at different levels of the position are zero.
Selector 402 is configurable to, and the first syntactic element and the second syntactic element are therefrom obtained for inverse symbolism device
Symbol subsequence in predetermined symbol type symbol, (it has predetermined the predetermined symbol type in symbol subsequence
Value of symbol and belong to identical subdivision) multiple prior search symbols come be directed to each symbol of predetermined symbol type selection
One in multiple contexts, or multiple prior search symbols of the predetermined symbol type in symbol sebolic addressing, it belongs to identical
Subdivision.According to above-mentioned specific embodiment, the first alternative be for coeff_abs_greater1 it is real, second
Alternative is real for coeff_abs_greater2.
Further, (wherein it is located and be more than according to the first binary syntax element for each position in one-dimensional path
One transformation series several levels) each position of display each transformation series several levels the 3rd syntactic element how many more than 1 can include it is whole
Numerical value syntactic element, i.e. coeff_abs_minus3, and inverse symbolism device 314 be configurable to use can be by control parameter
The mapping function of control is by the co-domain of the domain mapping of symbol sebolic addressing word to integer valued syntax elements, and if activation is efficient
Rate pattern, then it is configured to set each integer valued syntax member according to the integer valued syntax elements of the 3rd previous syntactic element
The control parameter of element, and be configured in the case of activating low-complexity pattern by be segmented it is constant in a manner of be configured and make
It must be arranged on and continuously keep constant in the subdivision of subsequence, wherein selector 402 is configurable in high efficiency mould
Under formula and low-complexity pattern, the symbol sebolic addressing word associated with equal-probability distribution for being mapped to integer valued syntax elements
Symbol select a predetermined entropy decoder (322).That is, even if it can be according to selected pattern against symbolism
Operated, but switch 400 and shown with dotted line 407.Inverse symbolism device 314 can keep control parameter during current clip
It is invariable such as global in time invariable, without the constant setting being segmented to control parameter.
Next the context modeling of complexity changeable will be described.
The same syntax of figs element for assessing the upper left neighbouring part of the derivation for context pattern index is a kind of conventional side
Method and often used in the case of HE, such as motion vector difference syntactic element.However, the assessment needs more cachings simultaneously
And do not allow direct coding syntactic element.Equally, can be to more available adjacent portions in order to reach higher coding efficiency
Divide and assessed.
In a preferred embodiment, all context modeling stages of neighbouring square or rectangle block or predicting unit
Syntactic element is assessed to be fixed in a context model.This equates disable adaptivity in the context model choice phase.
For this preferred embodiment, compared with CABAC current design, not to depending on the two of bit string after binarization
The context model selection of system position index is modified.In a further advantageous embodiment, except fixed context model it
Outside, for syntactic element using assessing neighbouring part, the context model of equally fixed different binary digit indexes.
Pay attention to, the description include for motion vector difference and the syntactic element relevant with encoding transform coefficients level binarization and
Context model selects.
In a preferred embodiment, only allow to assess left side neighbouring part.This can cause the buffer in process chain to reduce,
Reason is no longer to store last block or coding unit line.In further preferred embodiments, identical coding is only pointed to
Neighbouring part in unit is assessed.
In a preferred embodiment, all available neighbouring parts are assessed.For example, except upper left adjacent portion divides it
Outside, so that it may for property, upper left, upper right and lower-left neighbouring part can also be assessed.
That is, Figure 11 selector 402 is configurable to make a reservation for for relevant with the predetermined block of media data
Symbol, the symbol sebolic addressing relevant from more different adjacent blocks of media data is used in the case where activating high efficiency mode
Prior search symbol so as to select one in multiple contexts and according to and selected context-sensitive probabilistic model
Selected in entropy decoder 322.That is, adjacent block can be neighbouring on time and/or spatial domain.Spatially
Neighbouring block is visible, for example, in fig. 1-3.Then, under HE patterns, compared with LC patterns, selector 402 can
With the model selection that is carried out in response to mode switch 400 so as to based on the prior search symbol relevant with more adjacent blocks or
Syntactic element is adaptive contact, and thus reduces the storage overhead just described.
Next the coding of the reduction complexity to the motion vector difference according to embodiment is described.
In H.264/AVC video encoding and decoding standard, the motion vector associated with macro zone block passes through in current macrobiock
Motion vector and intermediary movements vector predictor between signal transmit poor (motion vector difference-mvd) and be transmitted.When
When CABAC is used as entropy coder, mvd coding is as follows.Integer value mvd is divided into absolute and is index portion.Absolute portion
Divide and utilize the group for blocking unitary and the 3rd rank index Columbus (prefix and suffix that are referred to as resulting bit string)
Close and carry out binarization.The binary digit relevant with blocking unary binarizations is encoded using context model, simultaneously
The binary digit relevant with index Columbus binarization is encoded in bypass mode, i.e. CABAC fixation probability is
0.5.The operation principle of unary binarizations is as follows.Assuming that mvd absolute integer value is n, then resulting bit string
It is made up of n ' 1 ' and a tail end ' 0 '.As example, it is assumed that n=4, then bit string is ' 11110 '.Just block unitary
For code, limiting value be present, and if the value over-limit condition, then bit string is formed by n+1 ' 1 '.For mvd
Situation, limiting value be equal to 9.This means if absolute value mvd is encoded equal to or more than 9,9 ' 1 ' are drawn, then two enter
Bit string processed with the binary prefix of index Columbus and suffix by forming.Block the process of the context modeling of unary part
It is as follows.For the first binary digit of bit string, if it is available, then fetching the mvd from upper left adjacent to macro zone block
Absolute value (if it is not, being inferred to the value for 0).If specific component (horizontal direction or vertical direction) and more than 2,
The second context model is then selected, if absolute value and more than 32, selects the 3rd context model, otherwise (absolute value and small
In 3), the first context model is selected.In addition, context model is different for each component.For binary digit
For second binary digit of string, the residue of unary part is used for using the 4th context model and the 5th context model
Binary digit.When mvd absolute values are equal to or more than 9, such as block all binary digits of unary part and be equal to ' 1 ', mvd
Absolute value and 9 difference are encoded using the 3rd rank index Columbus binary system in bypass mode.In last step,
Mvd label is encoded in bypass mode.
Defined in the current test model of high efficiency Video coding (HEVC) project using CABAC and be used as entropy code
Mvd newest coding techniques during device.In HEVC, resource block size is variable and the shape as defined in motion vector is referred to as predicting
Unit (PU).The PU sizes of upper left neighbouring part can have the other shapes and size in addition to current PU.Therefore, whenever
When related, the definition of upper left neighbouring part is now known as the upper left neighbouring part in the current PU upper left corner.For code book
For body, it can only be changed for the derivation of the first binary digit according to embodiment.Each neighbouring part can be independent
Assessed, without assess the MV from neighbouring part absolute value and.If the absolute value MV of neighbouring part can be used and is more than
16, then context model index can add up, the context model of identical quantity is produced thus directed towards the first binary digit, together
When residue MVD absolute values level and label coding with H.264/AVC in it is just the same.
In the above-mentioned technology encoded on mvd, it is necessary to encode up to 9 binary digits using context model, simultaneously
Mvd surplus value can be encoded together with label information under low-complexity bypass mode.Present embodiment describes reduction
Thus increase the quantity of bypass using the quantity of the binary digit of context model coding, and reduce upper needed for coding mvd
The hereafter technology of the quantity of model.Therefore, cutoff is down to 1 or 2 from 9.If this mean onlys that regulation, mvd absolute values are more than
Zero, if then the first binary digit is more than zero-sum one using context model to encode or provide mvd absolute values, first
Encoded with the second binary digit using context model, while surplus value is entered in bypass mode and/or using VLC codes
Row coding.Using VLC codes without using unitary or block unitary code as caused by binarization all binary digits using low multiple
Polygamy bypass mode encodes.In the case of PIPE, it is directly inserted into data flow and is inserted from data flow.Moreover, such as
If fruit has, then the more preferable context model of the first binary digit can be drawn using the different definition of upper left neighbouring part
Selection.
In a preferred embodiment, Exp- Golomb is used for the remainder of binarization MVD absolute value components.Therefore,
The order of Exp- Golomb is variable.The order of Exp- Golomb derives as follows.Drawing and encoding for the first binary system
After the context model of position, it follows that and after the index of coding context model, the index is used as index Columbus
The order of binarization part.In the preferred embodiment, the scope of the context model of the first binary digit is from 1 to 3, by
This obtains indexing 0-2, and it is used as the order of Exp- Golomb.The preferred embodiment can be used for HE situations.
During encoding MVD absolute values using in the alternative of the 2 above-mentioned technologies for being multiplied by 5 contexts, in order to
Encode 9 unitary code binarization binary digits, it is possible to use 14 context models (each component 7).For example, one
While first and second binary digits of first part are encoded using foregoing four different contexts, the 5th context
Available for the 3rd binary digit, the 6th context can be directed to the 4th binary digit and use, while the 5th to the 9th binary digit profit
Encoded with the 7th context.Therefore, in this case, in some instances it may even be possible to need 14 contexts, and only residue
Value can be encoded under low-complexity bypass mode.Reduce using context model coding binary digit quantity by
This increase bypass quantity, and reduce coding mvd needed for context model quantity technology be by cutoff for example from
9 are down to 1 or 2.If this mean onlys that regulation mvd, absolute value is more than zero, and the first binary digit is compiled using context model
If code or regulation mvd absolute values are more than zero-sum one, the first and second binary digits are compiled using each context model
Code, while surplus value is encoded using VLC codes.Using VLC codes, all binary digits as caused by binary system utilize low multiple
Polygamy bypass mode encodes.For PIPE, it is directly inserted into data flow and is inserted from data flow.In addition, proposed
Embodiment using another definition of upper left neighbouring part with draw be directed to the first binary digit more preferable context model select.
In addition, context modeling is modified, so that the context needed for the first or first and second binary digits
The quantity of model is reduced, so as to cause internal memory further to reduce.Equally, forbid entering for example above-mentioned neighbouring part of neighbouring part ratio
Row is assessed, and therefore saves on line buffer/memory needed for the mvd values of storage neighbouring part.Finally, can be with permission after side
The prefix binary digit (that is, the binary digit encoded using context model) of two components after the binary digit coding of road
The mode of coding split the coded sequence of component.
In a preferred embodiment, Exp- Golomb is used for the remainder of binarization mvd absolute value components.Therefore,
The order of Exp- Golomb is variable.The order of Exp- Golomb derives as follows.Drawing the context of the first binary digit
After model, it follows that after the index of context model, the index is used as the order of index Columbus binarization.
In the preferred embodiment, thus the scope of the context model of the first binary digit obtains indexing 0-2 from 1 to 3, its by with
Make the order of Exp- Golomb.The preferred embodiment can be used for HE situations and the quantity of context model is down to 6.In order to
The quantity of context model is reduced again and therefore saves on internal memory, and horizontal component and vertical component are in further preferred embodiments
In can share identical context model.In the case, it is only necessary to 3 context models.In addition, may only be by the left side
Neighbouring part account for being easy to being assessed in the another preferred embodiment of the present invention.In the preferred embodiment, no
Critical value can be modified (for example, only single critical value 16 can produce 0 or 1 index Columbus parameter or list
Individual critical value 32 can produce 0 or 2 index Columbus parameter).The preferred embodiment saves the row buffering needed for storage mvd
Device.In another preferred embodiment, critical value is changed and equal to 2 and 16.For this preferred embodiment, coding mvd is needed altogether
3 context models are wanted, and the scope of possible index Columbus parameter is 0-2.In further preferred embodiments, face
Dividing value is equal to 16 and 32.Again, described embodiment is applied to HE situations.
In the another preferred embodiment of the present invention, cutoff is down to 2 from 9.In the preferred embodiment, the one or two enters
Position processed and the second binary digit can be encoded using context model.The context model selection of first binary digit can
Modified in a manner of progress as in state of the art or more is stated described in preferred embodiment.For
Two binary digits, independent context model are selected as in state of the art.In further preferred embodiments
In, select the context model of the second binary digit by assessing the mvd of left side neighbouring part.Therefore, context model
Index is identical with the first binary digit, while available context model is different from the context model of the first binary digit.Always
6 context models (paying attention to, component shares context model) are needed altogether.Again, index Columbus parameter can depend on the
The selected context model index of one binary digit.In another preferred embodiment of the present invention, index Columbus parameter depends on
Indexed in the context model of the second binary digit.The described embodiment of the present invention can be used for HE situations.
In the another preferred embodiment of the present invention, the context model for two binary digits is fixed and not
Drawn by assessing left side or more neighbouring part.For the preferred embodiment, the sum of context model is equal to 2.At this
In the another preferred embodiment of invention, the first binary digit and the second binary digit share identical context model.Therefore,
Coding mvd only needs a context model.In the two preferred embodiments of the present invention, index Columbus parameter can be with
It is fixed and equal to 1.Described the preferred embodiments of the present invention are applied to both configurations of HE and LC.
In another preferred embodiment, context mould of the order of index Columbus part independently of the first binary digit
Type index is drawn.In this case, the absolute value and be used for for drawing that conventional context model H.264/AVC selects
The order of index Columbus part.The preferred embodiment can be used for HE situations.
In further preferred embodiments, the order of Exp- Golomb is fixed and is set to 0.Another preferred real
Apply in example, the order of Exp- Golomb is fixed and is set to 1.In a preferred embodiment, the order of Exp- Golomb is consolidated
It is set to 2.In a further embodiment, the order of Exp- Golomb is fixed as 3.In a further embodiment, index Columbus
The order of code is fixed according to current PU shape and size.The preferred embodiment proposed can be used for LC situations.Note
Meaning, considers the permanent order of index Columbus part using the binary digit of the reduction quantity with context model based coding.
In a preferred embodiment, neighbouring part is defined as follows.Received for above-mentioned PU, all PU for covering current PU
Enter and consider and using the PU with maximum MV.This is equally carried out for left side neighbouring part.All PU to covering current PU
Assessed and use the PU with maximum MV.In another preferred embodiment, the institute from the upper left border for covering current PU
The average absolute value motion vector value for having PU is used to draw the first binary digit.
For preferred embodiment proposed above, coded sequence can be changed by following.It must be directed to one by one
Horizontally and vertically specify mvd (vice versa).Therefore, it is necessary to two bit strings are encoded.In order to
The quantity of the pattern switching (switching i.e. between bypass mode and normal mode) of entropy code engine is minimized, can be entered to two
Position processed is encoded, and the binary digit is encoded using the context model for two components in the first step, is followed
The binary digit is the binary digit encoded in bypass mode in second step.Pay attention to, this be only resequence and
.
It note that as unitary or binary digit can also be indexed with each binary digit caused by blocking unitary binary system
The binary system of equivalent regular length of a mark represent that this mark specifies whether the value is more than current binary digit
Index.As example, the mvd binary cutoff of unitary that blocks is set to 2, thus obtain value 0,1,2 code word 0,
10、11.In the binary system of corresponding regular length of each binary digit index with a mark, binary digit index 0
One mark of (i.e. the first binary digit) specifies whether mvd absolute values are more than 0, and the two or two of binary digit index 1
One mark of system position specifies whether mvd absolute values are more than 1.When only encoding the second mark and the first mark is equal to 1, this
Identical code word 0,10,11 can be obtained.
Next the expression of the complexity changeable of the internal state of the probabilistic model according to embodiment will be retouched
State.
In HE-PIPE settings, the update probabilistic model after the internal state coding binary position with probabilistic model
Internal state.The state updated is tabled look-up by using the oldState of coding binary position and the State Transferring of value and drawn.Just
For CABAC, 63 different conditions can be presented in probabilistic model, wherein the mould that each state corresponds in section (0.0,0.5)
Type probability, these each states are used to realize two model probabilities.In addition to distributing to shape probability of state, 1.0 are also used
The probability is subtracted, and referred to as valMps mark is stored with the information that the probability whether is subtracted using probability or 1.0.This is total
126 states are obtained.In order to be used together this probabilistic model with PIPE Coded concepts, each needs in 126 states
One be mapped in effective PIPE encoders.In the current realization of PIPE encoders, this enters by using look-up table
OK.The example of this mapping is described in Table A.
In the following, it is described that the internal state of probabilistic model how can be represented to avoid using look-up table come will be interior
Portion's State Transferring is the embodiment of PIPE indexes.Some simple bit mask operations are only needed to carry out the inside shape from probabilistic model
PIPE indexes are extracted in state variable.The table of the novel complexity changeable of the internal state of estimate by design model in a manner of two level
Show.Application for forcing low-complexity operation, only uses the first order.Only describe pipeline index and for encoding or solving
Code-phase associates the mark valMps of binary digit.In the case of described PIPE entropy code schemes, the first order can be used
In 8 different model probabilities of differentiation.Therefore, the first order needs to identify for pipeIdx 3 positions and for valMps
Another position.Using the second level, each of the rough probable range of the first order is refined as supporting probability with higher resolution
The section of the number of smaller of presentation.More detailed present enables probability estimator more accurately to operate.In general, it is applicable
It is the coding application of higher RD performances in purpose.As example, the complexity of the internal state of PIPE probabilistic model is utilized
The expression of changeable is illustrated as follows：
First and second grades are stored in single 8 bit memory.Need 4 positions to store the first order, utilize most significant bit
On MPS value limit the index of PIPE indexes, 4 positions are used to store the second level in addition.In order to realize that CABAC probability is estimated
The behavior of gauge, each PIPE indexes have certain amount of allowed refinement index, and this depends on showing in PIPE indexes
Penetrate how many CABAC states.For example, for the mapping in Table A, the number of the CABAC states of each PIPE indexes is described in table B
Amount.
Table B：The quantity of the CABAC states of each PIPE indexes of the example of Table A
In the coding of binary digit or decoding process, PIPE indexes and valMps can be by using simple bit masks
Or displacement operation directly accesses.Low-complexity cataloged procedure only needs 4 positions of the first order, and high efficiency cataloged procedure can be with
Additionally the model renewal of CABAC probability estimators is carried out using 4 positions of the second level., can be with order to perform the renewal
Design point conversion lookup table, it is carried out and original table identical State Transferring, but want the complexity changeable of use state
Two level represents.Reset condition conversion table is multiplied by 63 elements by 2 and formed.For each input state, it includes two outputs
State.When being represented using complexity changeable, the size of state transition table is multiplied by 128 elements no more than 2, and this is rear sight
Very little acceptable increase.How many position the increase, which depends on, is used to represent refinement index and estimates for complete simulation CABAC probability
The behavior of gauge, it is therefore desirable to four positions.However, different probability estimator can be used, it can be to one group of CABAC shape of reduction
State works so as to each pipeline index for, it is allowed to state be no more than 8.Therefore, memory consumption can pass through tune
Matched suitable for representing the bit quantity of refinement index with the given complexity level of cataloged procedure.With 64 Probability States be present
The internal state of the model probability with CABAC of index is compared, and avoids being mapped to model probability specifically using tabling look-up
PIPE codes and need not further it change.
Next the renewal of the context model of the complexity changeable according to embodiment will be described.
For renewal context model, its Probability State indexes can be based on one or more previous codings two
System position updates.In HE-PIPE settings, the renewal is just carried out after each binary digit encodes or decodes.On the contrary,
In LC-PIPE settings, the renewal may be never carried out.
However, it is possible to carry out the renewal of context model in a manner of complexity changeable.That is, if more
The decision of new context model can be based on various aspects.For example, encoder, which is set, will not only update specific context model,
Such as syntactic element coeff_significant_flag context model, and always to update every other context model.
In other words, selector 402 can be configured to, for each symbol of multiple predetermined symbol types, according to it is each
The associated each probabilistic model of predetermined symbol is selected in entropy decoder 322 so that the quantity of predetermined symbol type is low
It is lower than under high efficiency mode under complexity model.
In addition, the standard for controlling whether to update context model can be for example the size of bit stream bag, solve so far
The quantity of the binary digit of code, or renewal are only entered the two of the specific fixed qty or variable number of coding context model
Carried out after position processed.
In the scheme for deciding whether to update context model, it is possible to achieve the context model of complexity changeable is more
Newly.This allows to increase or decrease part of the binary digit in the bit stream of context model renewal is carried out for it.Context
The quantity of model modification is higher, and code efficiency is better and computational complexity is higher.Therefore, described side can be utilized
Case come realize the context model of complexity changeable update.
In a preferred embodiment, context model renewal, syntactic element are carried out to the binary digit of all syntactic elements
Except coeff_significant_flag, coeff_abs_greater1 and coeff_abs_greater2.
In further preferred embodiments, only to syntactic element coeff_significant_flag, coeff_abs_
Greater1 and coeff_abs_greater2 binary digit carries out context model renewal.
In further preferred embodiments, all context models are carried out up and down when starting fragment coding or decoding
Literary model modification.After the transform blockiis of specific predefined quantity is handled, until reaching the end of fragment just to all upper and lower
Literary model carries out context model renewal.
For example, selector 402 is configurable to, for the symbol of predetermined symbol type, according to predetermined symbol type phase
Together with the probabilistic model of the association probabilistic model associated with updating or in the case where not updating associated probabilistic model
Selected in entropy decoder 322 so that in the symbol sebolic addressing of the selection and renewal of the symbol for carrying out predetermined symbol type thereon
Understanding the stage length it is shorter than under high efficiency mode under low-complexity pattern.
Further preferred embodiments are identical with previously described preferred embodiment, but the further preferred embodiments are with one
Kind mode is represented using the complexity changeable of the internal state of context model so that a table is stored with all contexts
" Part I " (valMps and the pipeIdx) of model, second table are stored with " Part II " of all context models
(refineIdx).(such as it is preferable to carry out when forbidding and carrying out context model renewal to all context models previously
Described in example), the table for being stored with " Part II " is no longer required and therefore can be dropped.
Next the renewal of the context model of the binary bit sequence according to embodiment will be described.
LC-PIPE configuration in, type coeff_significant_flag, coeff_abs_greater1 and
The binary digit of coeff_abs_greater2 syntactic element is grouped into multiple subsets.It is single upper and lower for each subset
Model is used to encode its binary digit.In this case, context model renewal can encode the fixed qty of the sequence
Binary digit after carry out.Hereinafter this represents the renewal of polybinary position.However, the renewal using only last with encoding
Binary digit and the internal state of context model renewal it is different.For example, for each binary digit of coding, carry out
One context model updates step.
Hereinafter, the coding of the exemplary subset to being made up of 8 binary digits gives example.Alphabetical ' b ' is represented
Binary digit is decoded, alphabetical ' u ' represents renewal context model.In the case of LC-PIPE, without context model more
Lower progress binary digit decoding of news：
b b b b b b b b
In the case of HE-PIPE, after each binary digit is decoded, context model renewal is carried out：
b u b u b u b u b u b u b u b u
In order to somewhat reduce complexity, context model renewal can be carried out after a series of binary digits (in the reality
It is after every 4 binary digits in example, carries out the renewal of this 4 binary digits)：
b b b b u u u u b b b b u u u u
That is, selector 402 is configurable to, for the symbol of predetermined symbol type, according to predetermined symbol class
Together with the associated probabilistic model of the type probabilistic model associated with updating or do not updating the situation of associated probabilistic model
Under selected in entropy decoder 322 so that the symbol of predetermined symbol type selection with renewal by its progress frequency exist
It is lower than under high efficiency mode under low-complexity pattern.
In this case, after 4 binary digits are decoded, 4 renewal steps are entered based on 4 two decoded just now
Position processed.Pay attention to, this four renewal steps can be carried out by using special look-up table is searched in a single step.At four
After tradition renewal step, the look-up table is each for each possible internal state of 4 binary digits and context model
May the resulting new state of combination storage.
Under some pattern, polybinary position is updated for syntactic element coeff_significant_flag.For institute
There is the binary digit of other syntactic elements, updated without using context model.Compiled before polybinary position renewal step is carried out
The quantity of the binary digit of code is set to n.When the quantity of the binary digit of setting is not divided exactly by n, 1 to n-1 binary digit exists
The end of subset is retained in after last polybinary position renewal.For each of these binary digits, traditional single two
The renewal of system position is carried out after all these binary digits are encoded.Quantity n may be greater than 1 positive count.Another
Pattern can be equal to preceding mode, for coeff_significant_flag, coeff_abs_greater1 and coeff_
Abs_greater2 (instead of only there was only coeff_significant_flag) any combination carries out polybinary position renewal
Except.Therefore, the pattern can be more more complicated than other patterns.Every other syntactic element is (wherein without using polybinary position more
Two disjoint subsets can newly) be divided into (wherein for one of subset, to update using single binary digit, for other
For subset, updated without using context model).Any possible disjoint subset is all effectively (including vacuous subset).
In an alternative embodiment, the renewal of polybinary position is only based on compiling immediately before polybinary position updates step
The last m binary digit of code.M can be less than n random natural number.Therefore, following decoding can be carried out：
b b b b u u b b b b u u b b b b u u b b b b…
Wherein n=4 and m=2.
That is, selector 402 is configurable to, for the symbol of predetermined symbol type, according to predetermined symbol class
The phase every n symbols of the associated probabilistic model of type and the m based on the predetermined symbol type predefined types of symbol renewal recently
The probabilistic model of association is selected in entropy decoder 322 together so that ratio n/m is under low-complexity pattern than high
It is high under efficiency mode.
, will be as described above for syntactic element coeff_significant_flag in further preferred embodiments
The context modeling scheme that local template is used for HE-PIPE configurations can be used for context model distributing to syntactic element
Binary digit.However, for these binary digits, updated without using context model.
Further, selector 402 is configurable to, for the symbol of predetermined symbol type, according to the more of symbol sebolic addressing
The symbol of individual prior search select one in multiple contexts and according to selected context-sensitive probabilistic model in entropy
Selected in decoder 322 so that the quantity of context and/or the quantity of the symbol of prior search, in low-complexity pattern
It is lower fewer than under high efficiency mode.
Initialized using the probabilistic model of 8 initialization values
This section describe the complexity for the probabilistic model that two 8 place values are replaced using so-called 8 initialization values to increase
The initialization procedure of the internal state subtracted, as state of the art video encoding standard H.265/AVC in situation one
Sample.It is made up of two parts, and the two parts can be with the initialization value for the probabilistic model in CABAC H.264/AVC
To being compared.The two parts represent that two parameters of linear equation to calculate the original state of probabilistic model, represent to come from
QP particular probability (for example, being in PIPE indexes form)：
Part I describes slope and using being related to the interior of the quantization parameter (QP) that is used during coding or decoding
The dependence of portion's state.
Part II is defined on given QP PIPE indexes and valMps.
Two different modes can be used for using given initialization value come initialization probability model.First mode is expressed as
Initialization independent of QP.Only the PIPE indexes defined in the Part II by initialization value and valMps are used for all QP.
This equates the situation that slope is equal to 0.Second mode is expressed as relying on QP initialization, and it is in addition using the of initialization value
The slope of a part indexes to change PIPE indexes and define refinement.The two of 8 initialization values are shown partially as follows：
It is made up of two 4 bit positions.Part I include point to storage in an array 16 it is different it is predefined tiltedly
The index of 1 in rate.Predefined slope is by 7 negative slopes (slope indexes 0-6), (slope indexes a null slope
7) and 8 positive slopes (slope indexes 8-15) form.Table C describes slope.
Table C：
All values carry out scale to avoid using floating-point operation with 256 coefficient.Part II is to embody probability interval p
The PIPE indexes of the increasing probability of valMps=1 between=0 and p=1.In other words, PIPE encoders n must with than
Model probability high PIPE encoders n-1 is operated.For each probabilistic model, a PIPE probability index is available
And identify that its probability interval includes p for QP=26ValMPs=1Probability PIPE encoders.
Table D：The Part II of initialization value is mapped to PIPE encoders and valMps：UR=unitary is to this code of Lay
(unary-to-rice-code), tri- binary system bit codes of TB=, BP=binary digit pipeline codes, EP=equiprobability are (uncoded
)
QP and 8 initialization value is needed come the simple linear equation by calculating y=m* (QP-QPref)+256*b forms
To calculate the initialization of the internal state of probabilistic model.Note：M definition utilizes slope index (Part I of 8 initialization values)
The slope obtained from table C, b represent the PIPE encoders (Part II of 8 initialization values during QPref=26：" PIPE is general
Rate indexes ").Then, valMPS is 1 and if y is more than 2047, then pipeIdx is equal to (y-2048)>>8.Otherwise valMPS
Equal to 0 and pipeIdx is equal to (2047-y)>>8.If valMPS is equal to 1, refinement index is equal to (((y-2048) ＆
255)*numStates)>>8.Otherwise, refinement index is equal to (((2047-y) ＆255) * numStates)>>8.In both feelings
Under condition, numStates is equal to the quantity of the CABAC states of the pipeIdx as described in table B.
Such scheme can not only combine PIPE encoders and use, and can be combined with CABAC schemes mentioned above
Use.In the case of in the absence of PIPE, each PIPE Idx CABAC states, i.e., carried out therebetween during probability updating
The Probability State (pState_current [bin]) (that is, pState_current [bin] each most significant bit) of State Transferring
Quantity then simply realized indeed according to QP CABAC states piecewise linear interpolation one group of parameter.In addition, in parameter
In the case that numStates uses identical value to all PIPE Idx, piecewise linear interpolation almost can be with disabled.For example,
NumStates is arranged to 8 for all situations and always meets 16*8 state of generation together, and the calculating for refining index is directed to
ValMPS is reduced to ((y-2048) ＆255) equal to 1>>5, ((2047-y) ＆255) is reduced to equal to 0 for valMPS>>5.It is right
In such case, using valMPS, PIPE idx and idx expression will be refined map back original CABAC institutes H.264/AVC
The expression used is very simple.CABAC states are with (PIPE Idx<<3)+refinement Idx form provides.Enter below for Figure 16
One step describes this respect.
Unless the slope of 8 initialization values be equal to zero or unless QP be equal to 26, otherwise must by using linear equation profit
Internal state is calculated with coding or the QP of decoding process.With regard to slope be equal to zero or current encoding procedure QP equal to 26
For situation, the Part II of 8 initialization values is used directly for the internal state of initialization probability model.Otherwise, thus
The fractional part of caused internal state can be further used in high efficiency coding application by specific PIPE encoders
Linear interpolation between limitation indexes to determine to refine.In the preferred embodiment, linear interpolation by simply will to work as
The sum of available refinement index is multiplied by the fractional part and is mapped to the result immediate for preceding PIPE encoders
Refinement is indexed to perform.
The process of the initialization of the internal state of probabilistic model can change for the quantity of PIPE probability Index Status
Become.Specifically, can avoid occurring twice using PIPE encoders E1 equiprobability pattern, i.e. can avoid different using two
PIPE indexes 1 are still 0 to distinguish MPS to be, as follows.Again, the process can be called during starting to parse fragment data, and
And the input of the process is 8 initialization values as described in table E, it is for example in each context model to be initialised
It is transmitted in bit stream.
Table E：8 of the initValue of probabilistic model are set
First 4 define slope and index and retrieved by mask bit b4-b7.Indexed for each slope, in table F
In specify and show slope (m).
Table F：SlopeIdx variable m value
| slopeIdx | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| m | -239 | -143 | -85 | -51 | -31 | -19 | -11 | 0 |
| slopeIdx | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 |
| m | 11 | 19 | 31 | 51 | 85 | 143 | 239 | 399 |
Position b0-b3, latter 4 of 8 initialization values, identify probIdx and describe the probability at predefined QP.probIdx
0 indicated value is the maximum probability of 0 symbol, and respectively, the indicated values of probIdx 14 are the maximum probability of 1 symbol.Table G pins
PipeCoder and its valMps corresponding to being shown to each probIdx.
Table G：By rear 4 bit mapping of initialization value to PIPE encoders and valMps：UR=unitary is to this code of Lay, TB=
Three binary system bit codes, BP=binary digit pipeline codes, EP=equiprobability (uncoded)
Using the two values, the calculating of internal state is carried out by using such as y=m*x+256*b linear equation, its
Middle m represents slope, and x represents that the QP of current clip, the probIdx shown in b from following description are drawn.All values during this
Scale is carried out with 256 coefficient to avoid using floating-point operation.The output (y) of the process represents probabilistic model at current QP
Internal state and be stored in 8 bit memories.As shown in G, internal state is by valMPs, pipeIdx and refineIdx group
Into.
Table H：The internal state of probabilistic model is set
RefineIdx and pipeIdx distribution is similar to the internal state (pStateCtx) of CABAC probabilistic models and is in
In present H.
Table I：PipeIdx, refineIdx and pStateCtx distribution
In a preferred embodiment, probIdx is defined at QP26.Based on 8 initialization values, to probabilistic model
Internal state (valMps, pipeIdx and refineIdx) is handled, as described in following pseudo-code：
As shown in pseudo-code, refineIdx is quantified as pair by the linear interpolation between pipeIdx section and by result
RefineIdx is answered to be calculated.The offset represents the total quantity of the refineIdx relative to each pipeIdx.
FullCtxState/256 section [7,8) dimidiation.Section [7,7.5) is mapped to pipeIdx=0 and valMps=
0, and section [7.5,8) it is mapped to pipeIdx=0 and valMps=1.Figure 16 describes the process for drawing internal state and shown
Show fullCtxState/256 to pStateCtx mapping.
Pay attention to, slope indicates probIdx and QP dependence.If the slopeIdx of 8 initialization values is equal to 7,
The internal state of resulting probabilistic model is identical for all fragment QP, therefore the initialization of internal state
Current QP of the journey independently of fragment.
Initialized that is, selector 402 can index to pipeline, the pipeline is indexed for utilizing instruction
Used quantization step QP syntactic element come decoding data stream with lower part such as entirely stream or next fragment, so as to
Quantify the data of the part, for example be included in be total to using the syntactic element as the two patterns LC and HE
The transformation series several levels of the index of table.The table such as table D can include each sign pattern, respectively refer to QPref pipeline rope
Draw or other data of each sign pattern.According to the actual QP of current portions, selector can be utilized by actual QP and QP
Each table clause a of index itself calculates pipeline index value, such as by the way that (QP-QPref) is multiplied by into a to be calculated.LC moulds
Unique difference of formula and HE patterns is：Compared with HE patterns, in the case of LC, in terms of selector is only come by relatively low precision
Calculate result.Selector for example can only use the integer part of result of calculation.Under HE patterns, the remainder of degree of precision, such as
Fractional part, for one in the available refinement index for each pipeline index for selecting to indicate such as lower accuracy or integer part
It is individual.Using refinement index (may less being used under LC patterns) so as to by using table row mentioned above under HE patterns
(table walk) is walked to carry out probability adaptation.When the available index for indexing current pipeline is stayed in compared with high threshold, connect
Get off by the way that subdivision index is minimized, select higher pipeline index.When the available index for indexing current pipeline stay in compared with
When at lower limit, the available maximum for new pipeline index is arrived followed by refinement index is maximized, is selected next
Individual relatively low pipeline index.Pipeline indexes and Probability State is defined together with refinement index, but for the selection in the stream of part,
Selector indexes using only pipeline.Refinement index is only used for carrying out tracking probability more closely or with more preferable precision.
However, discussed above also illustrate that complexity changeable can be independently of Fig. 7-10 PIPE Coded concepts or profit
Realized with the CABAC of decoder as shown in figure 12.Figure 12 decoder is used for the data that decoding media data is encoded into
Stream 601, and be configured to be swashed according to data flow 601 including mode switch 600 and inverse symbolism device 602, the mode switch 600
Low-complexity pattern or high efficiency mode living, this is configured to inverse symbolism directly or by entropy decoding (example against symbolism device 602
As obtained from data flow 601) symbol sebolic addressing 603 is using the controllable mapping function of control parameter to obtain integer value language
Method element 604, so as to by the co-domain of the domain mapping of symbol sebolic addressing word to integer valued syntax elements.Reconstructor 605 is configured to
Integer valued syntax elements carry out re-constructing media data 606.Inverse symbolism device 602 is configured to carry out inverse symbolism so that high in activation
Control parameter is changed according to data flow with first rate in the case of efficiency mode, and in the situation of activation low-complexity pattern
Lower control parameter invariable (regardless of data flow) or according to data flow with less than the second speed of first rate change,
As illustrated with arrow 607.For example, control parameter can be according to the sign modification of previous inverse symbolism.
In above-described embodiment some utilize Figure 12 aspect.Syntactic element coeff_abs_minus3 in sequence 327
It is binarized with MVD for example in inverse symbolism device 314 according to the pattern of 407 selected instructions, and reconstructor 605 makes
Rebuild with these syntactic elements.Obviously, it is easy to combine in terms of the two of Figure 11 and Figure 19, but Figure 12 aspect may be used also
To be combined with other coding environments.
Encoded for example, with reference to motion vector difference illustrated above.Inverse symbolism device 602 can be arranged such that mapping letter
Number blocks unitary code using unitary code is blocked in the domain of the integer valued syntax elements less than cutoff, and in cutoff
The prefix of form and in VLC code words suffix combination first interval in, be more than and equal to cutoff integer value language
Mapped in the second interval in the domain of method element, wherein decoder can include entropy decoder 608, and the entropy decoder 608 is matched somebody with somebody
It is set to multiple first binary digits and profit for being drawn using the entropy decoding for changing probability Estimation from data flow 601 and blocking unitary code
Multiple second binary digits of VLC code words are drawn with constant equiprobability bypass mode.Under HE patterns, entropy code ratio is in arrow
It is more complicated in LC codings shown in first 609.That is, can be with application context adaptivity and/or probability under HE patterns
Suppressed adaptively and under LC patterns, or, in other words, scale can be carried out to complexity, as described above for each reality
Apply what example was stated.
The encoder being used for by media data encoding for data flow of suitable Figure 11 decoder is shown in Figure 13.Should
Encoder can include inserter 500, be configured to signal notice low-complexity pattern or high efficiency mode in data flow 501
Activation, constructor 504, it is syntactic element sequence 506 to be configured to the precoding of media data 505, symbolism device 507, configuration
For the symbol of syntactic element sequence 506 is turned into symbol sebolic addressing 508, multiple entropy coders 310, wherein each be configured to part
Symbol sebolic addressing is converted to the code word of data flow, and selector 502, is configured to turn each symbol of symbol sebolic addressing 508
Selected in multiple entropy coders 310 one is sent to, wherein selector 502 is configured to according to the low complexity as shown in arrow 511
Activated in sexual norm and high efficiency mode one is selected.Interleaver 510, which can optionally be set, carrys out interleaving coder
310 code word.
Being used for the encoder that media data encoding is data flow for suitable Figure 12 decoder is shown in Figure 14, its
Including inserter 700, the activation of signal notice low-complexity pattern or high efficiency mode in data flow 701, construction are configured to
Device 704, it is to include the syntactic element sequence 706 of integer valued syntax elements, symbolism to be configured to the precoding of media data 705
Device 707, it is configured to using the controllable mapping function of control parameter come symbolism integer valued syntax elements so as to by integer value language
The domain mapping of method element to symbol sebolic addressing word co-domain, wherein symbolism device 707 be configured to carry out symbolism so that activating
Control parameter is changed according to data flow with first rate in the case of high efficiency mode, and in the feelings of activation low-complexity pattern
Under condition control parameter invariable (regardless of data flow) or according to data flow with less than the second speed of first rate become
Change, as may be indicated by arrow 708.Symbolism result is encoded to data flow 701.
Again, it is notable that Figure 14 embodiment is easily converted to context-adaptive two mentioned above and entered
Arithmetic coding/decoding embodiment processed：The compressible into direct output stream 401 of selector 509 and entropy coder 310 simultaneously selects to work as
It is preceding will be from the context adaptive binary arithmetic coding device of the context for the binary digit that data flow is drawn.This is to context
It is especially real for adaptive and/or probability adaptation.During low-complexity pattern, function/adaptivity can
Close, or design is lighter.
Briefly touch upon above, some elaborations of above-described embodiment can need not be directed to according to alternative embodiment
Mode switching capabilities.In order to clarify this point, reference picture 16, it is summarized to foregoing description, as long as mode switching capabilities
Deletion make a distinction Figure 16 embodiment and above-described embodiment.Moreover, following description will display and the phase such as H.264
The advantages of Billy is drawn with less accurate slope and offset parameter to initialize the probabilistic estimated value of context.
Specifically, Figure 16 is shown for the decoder from the decoding video 405 of data flow 401, using horizontal component and is hung down
The horizontal component of motion vector difference and vertical component are encoded into the data flow by the binarization of straight component, in horizontal component
In the first interval in the domain less than cutoff of vertical component, the binarization is respectively equal to horizontal component and vertical point
Amount blocks unitary code, and the combination of prefix is in the form of blocking unitary code.Respectively in the finger of horizontal component and vertical component
The cutoff and suffix of number Columbus code forms, respectively, in being more than and equal to cutoff for horizontal component and vertical component
In the second interval in domain, wherein cutoff is 2 and Exp- Golomb has order 1.Decoder includes：Entropy decoder 409,
It is configured to, for the horizontal component and vertical component of motion vector difference, using context adaptive binary entropy decoding from data
Stream, which is drawn, blocks unitary code, and each binary digit position for blocking unitary code has a proper context, and it is to motion vector
It is common for the horizontal component and vertical component of difference, and Exp- Golomb is drawn using constant equiprobability bypass mode
To obtain the binarization of motion vector difference.More precisely, as described above, entropy decoder 409 is configurable to utilize two
(than CABAC schemes as mentioned above, or binary system PIPE is decoded system entropy decoding, i.e., using being related to some parallel work-flows
The construction of entropy decoder 322 and each selector/distributor) draw from data flow 401 binary digit 326 of binarization
Quantity.Inverse symbolism device 314 is against the binarization of binarization motion vector difference syntactic element to obtain the water of motion vector difference
The integer value of the amount of dividing equally and vertical component, and horizontal component and vertical component of the reconstructor 404 based on motion vector difference
Integer value rebuilds video.
In order to which such case is explained in more detail, referring briefly to Figure 18.800 typically illustrate a motion vector difference,
That is, the vector of the prediction residual between predicted motion vector and reality/reconstructed motion vector is represented.It also show horizontal component
With vertical component 802x and 802y.Can also be with location of pixels, i.e. pel spacing, or sub-pixel location such as pel spacing
Half or its a quarter etc. are transmitted for unit.Horizontal component and vertical component 802x,yIt is integer value.They
Domain is from zero to infinity.Index value can be handled individually, therefore here without further considering.In other words, originally
The size for focusing on motion vector difference 802x, y of the description of text general introduction.The domain is shown at 804.The right hand of axle 804 in domain
Side, Figure 19 show the associated binarization of the probable value of the component 802x, y on each other with vertical arrangement, each possibility
Value be mapped to (binarization is arrived) described binarization.Unitary code is blocked as can be seen that only occurring below 2 in cutoff
806, and binarization also has the finger of the order 808 since equal to or more than the probable value of cutoff 2 as suffix
Number Columbus's code, binarization is carried out to continue to subtract cutoff more than 1 remainder of integer value.Enter for all two
Position processed, only provide two contexts：One the first binary system for being used for horizontal component and vertical component 802x, y binarization
Position position, another is used for horizontal component and vertical component 802x, y the second binary digit position for blocking unitary code 806.
For the binary digit position of Exp- Golomb 808, the entropy-decoded device 409 of equiprobability bypass mode uses.That is,
Assuming that the two binary bit value equiprobability occur.The probability Estimation of these binary digits is fixed.In contrast to this, solving
Two context-sensitive probability of the constantly binary digit of blocking unitary code 806 of the adjustment with referring to just now are estimated during code
Meter.
Appointed in more detailed description entropy decoder 409 according to what how foregoing description was implemented to be referred to just now
Before business, the present possibility for focusing on reconstructor 404 of the description realizes that the reconstructor 404 uses motion vector difference 800
With such as inverse symbolism device 314 by the binary digit of binarization code 106 and 108 again and the integer value that obtains, again two
Enter inhibition and generation to show using arrow 810 in figure 18.Specifically, reconstructor 404 from data flow 401 as set forth above, it is possible to retrieve
It is related to the information that the picture that will currently rebuild is subdivided into block, it is pre- that at least some in the block need to carry out motion compensation
Survey.Figure 19 the picture typically to rebuild is shown at 820 and shown at 822 picture 120 segment block, just now
The subdivision motion compensated prediction referred to is used to predict image content therein for the block.As described in being directed to Fig. 2A -2C,
Subdivision to block 122 is different with probability for size.In order to avoid transmitting each motion vector difference of these blocks 122
800, reconstructor 404 can utilize merges concept, data flow is transmitted in addition to subdivided information in addition accordingly, or in the absence of
In the case of subdivided information, other pooling informations in addition to the situation of fixed subdivision.Pooling information signal notifies reconstructor
Which of 404 blocks 822 block forms merging group.Using this measure, some can be moved for reconstructor 404
Phasor difference 800 is applied to the whole merging group of block 822.Naturally, in coding side, transmission pooling information need to open in subdividing transmission
Pin (if present), pooling information transport overhead and with merging group size increase and reduce motion vector difference transmit
Traded off between expense.On the other hand, the motion vector of the merging group can be reduced by increasing the number of blocks of each merging group
The adaptation of actual demand of the difference to the independent blocks of each merging group, the motion of the motion vector difference of these blocks is thus caused to be mended
Repay and predict that less accurate and so that the higher transmission expense of communicating predicted residual error necessitates in the form of transformation series several levels etc..Phase
Ying Di, found in an appropriate manner on coding side compromise.However, in any case, merge the motion that concept makes merging group
Phasor difference shows less interior volume correlation.For example, with reference to Figure 19, it is shown to some merging group by beating shade
Member relation.Obviously, the actual motion of the image content in these blocks is so similar so that coding side determines to merge each area
Block.It is however, relatively low with the correlation of the motion of the image content in other merging groups.Correspondingly, unitary is blocked relative to each
The limitation that a context is used only in the binary digit of code 806 will not have a negative impact to entropy-coding efficiency, because merging general
Thought has fully adapted to the interior volume correlation between the motion of adjacent picture content.Context can only be based on binary system
Position is motion vector difference component 802x,yWith due to cutoff be equal to 2 caused by two enter for 1 or 2 binary digit position
The fact that a part for inhibition and generation, is selected.Correspondingly, other decoded binary digit/syntactic element/mvd components
802x,yContext selection is not influenceed.
Similarly, reconstructor 404 is configurable to assume to predict that concept will be further by fortune to reduce by using more
The information content that dynamic vector poor (space and/or the time prediction that exceed motion vector) transmits.It is general according to more hypothesis predictions
Read, motion vector predictor list is generated first for each block or merging group, then on being actually subjected to be used for predicted motion
Clearly or implicitly it is transmitted in the traffic flow information of the index of the fallout predictor of phasor difference.For example, with reference to non-in Figure 20
Shade block 122.Reconstructor 404 can be come to provide different fallout predictors for the motion vector of the block in the following manner, than
Such as from the left side, from top, or from combination of the two etc. from spatially predicted motion vector, and according to the previous of video
The motion vector of the co-located part of decoding picture and the predicted motion in time that is further combined of foregoing fallout predictor are sweared
Amount.These fallout predictors by encode side it is foreseeable it is predictable in a manner of classified by reconstructor 404.Some information are transferred into
This one end and reconstructed device in data flow are utilized.That is, comprising some hint in data flow, it is to close that this, which is implied,
In which ordered list from fallout predictor fallout predictor actually application make the block motion vector prediction its.The rope
Draw clearly to be transmitted for the block in data flow.However, the index can also be predicted first, then only
It is the prediction of the index of transmission.There is also other probability.Under any circumstance, the prediction scheme referred to just now can be to current
The motion vector of block carries out point-device prediction, and therefore reduces the information content demand being imposed in motion vector difference.
Correspondingly, context-adaptive entropy code is only down to 2 (such as pins to the limitation for two binary digits for blocking unitary code, cutoff
To described in Figure 18), and the order of selection Exp- Golomb is 1, and code efficiency will not be had a negative impact, because fortune
Dynamic vector difference shows frequency histogram due to high forecasting efficiency, according to frequency histogram motion vector difference component 802x,
Y high value is not often accessed.Omit any difference between horizontal component and vertical component and be adapted to efficiently prediction, because
The prediction is intended in the high both direction of precision of prediction often equal operation.
It is important to note that in the foregoing description, the whole details for being provided with Fig. 1-15 also can be exchanged into institute in Figure 16
The entity shown, for example only relate to the function of inverse symbolism device 314, reconstructor 404 and entropy decoder 409.But, it is
For the sake of complete, below the part in these details will be summarized again.
Referring to Figure 20 to be better understood from the prediction scheme summarized just now.As just mentioned, constructor 404 can obtain
The different fallout predictors of the current merging group of current block 822 or block are obtained, wherein these fallout predictor solid line vectors 824 are shown.
Fallout predictor can be obtained by space and/or time prediction, wherein, additionally, arithmetic mean of instantaneous value computing etc. can be used, is made
Obtaining independent fallout predictor can be obtained in a manner of these fallout predictors are associated with each other by reconstructor 404.Independently of acquisition vector
These orders of fallout predictor 126 are turned to or are categorized as ordered list by 826 this mode, reconstructor 404.This is used in Figure 20
Digital 1-4 is shown.Assorting process can be preferably uniquely determined, so that encoder and decoder can be with simultaneously operating.Then,
Just now the index referred to can be directed to current block or merging group clearly or is implicitly obtained by reconstructor 404 from data flow
.For example, it may be possible to chosen second fallout predictor " 2 ", and that motion vector difference 800 is added into this is selected for reconstructor 404
Fallout predictor 126, the motion vector 128 finally rebuild thus is produced, the motion vector 128 is subsequently used for pre- by motion compensation
Survey to predict the content of current block/merging group.For merging group, reconstructor 404 may include carrying for the block of merging group
The further motion vector difference supplied, so that the independent block further directed to merging group carrys out refined motion vector 128.
Therefore, continue to describe the realization of the entity shown in Figure 16, entropy decoder 409 is configurable to utilize
Binary arithmetic decoding or binary system PIPE codings draw from data flow 401 and block unitary code 806.It is general to the two above
Thought is described.Further, entropy decoder 409 is configurable to be used to different contexts block the two of unitary code 806
Individual binary digit position, or alternatively, same context is used for the two binary digits.Entropy decoder 409 can be configured to
Carry out Probability State renewal.Entropy decoder 409 by for currently from block binary digit that unitary code 806 draws from pin
To the context-sensitive current probability State Transferring that the binary digit that currently draws selects into two depending on currently drawing
The new Probability State of system position updates to perform Probability State.Referring to upper table Next_State_LPS and Next_State_MPS,
The table is searched relative to the operation performed in addition to other steps 0-5 listed above by entropy decoder.Discussed above
In, current probability state is referred to by pState_current.It is defined for each context interested.Entropy decoder
409 are configurable in the following manner come to currently to carry out binarization from the binary digit that unitary code 806 draws is blocked
Arithmetic decoding：The current probability interval width value (i.e. R) in quantization means current probability section, to obtain probability interval index q_
Index, and indexed and depending on the context phase with being selected for the binary digit currently to be drawn by using probability interval
The Probability State index (i.e. p_state) of the current probability state of association, it is that the table clause in table clause is indexed to carry out area
Between segment, using obtain by current probability interval subdivision as two partial sections.In the embodiment of above-outlined, these parts
Section is associated with the symbol of maximum probability and minimum probability.As described above, entropy decoder 409 is configurable to by crawl 8
2 or 3 most significant bits that position represents etc. and quantization current probability interval width value, using wide for current probability section
Angle value R 8 expressions.Entropy decoder 409 can be further configured to be based on the inside from current probability section (i.e. V)
Bias state value selects from two partial sections, update probability interval width value R and bias state value, and selected by utilization
Partial section is inferred to the value for the binary digit currently to be drawn, and to the continuity including the reading position from data flow 401
Update probability interval width value R and bias state value V are standardized again.Entropy decoder 409 is for example configurable to, and is led to
Cross etc. and to be divided to current probability interval width value to come current probability interval subdivision to from index brother as two partial sections to obtain
The binary digit of human relations cloth code carries out binary arithmetic decoding.The decile corresponds to probabilistic estimated value, and the probabilistic estimated value is fixed
And equal to 0.5.This can be realized by simple displacement.Entropy decoder can be further configured to, for each motion
Phasor difference, before the horizontal component of each motion vector difference and the Exp- Golomb of vertical component, drawn from data flow 401
The horizontal component of each motion vector difference and vertical component block unitary code.Using this measure, entropy decoder 409 can profit
Form a series of binary digits together with more binary digits, probabilistic estimated value be to this it is fixed, as 0.5.This can be with
Accelerate the process of entropy decoding.On the other hand, the water that entropy decoder 409 may preferably by drawing a motion vector difference first
The amount of dividing equally and vertical component, then only continue to draw horizontal component and the vertical component of next motion vector difference to keep moving
Order between phasor difference.Using this measure, the memory requirements for imposing on decoding entity, i.e. Figure 16 decoder is reduced, because
It can continue immediately against binarization motion vector difference, without waiting to other motion vector difference for inverse symbolism device 314
It is scanned.Such case is realized by context selection：Since it is desired that the binary digit position for each code 806
Exactly only have a context can use, so the correlation in space need not be checked.
Reconstructor 404 be configurable to as described above from the horizontal component of space and/or time upper predicted motion vector and
Vertical component is to obtain the fallout predictor 126 of the horizontal component of motion vector and vertical component and by using motion vector difference
Horizontal component and vertical component refinement fallout predictor 826 (for example being added to each fallout predictor only by by motion vector difference) come
The horizontal component and vertical component of reconstructed motion vector.
Further, reconstructor 404 is configurable to come in a different manner the horizontal component of predicted motion vector and hung down
Straight component is arranged to obtain the sorted lists of the fallout predictor of the horizontal component of motion vector and vertical component from data flow
Table index and by using motion vector difference horizontal component and vertical component refinement list index point to list fallout predictor
Carry out the horizontal component and vertical component of reconstructed motion vector.
Further, it is as already described above, reconstructor 404 be configurable to using motion compensated prediction by with
Horizontal component and the vertical component 802x, y of the spatial granularity application motion vector that block limits are subdivided into by video pictures to weigh
Video is built, wherein reconstructor 404 can use the merging syntactic element being present in data flow 401 so as to which block is grouped into
Merging group is simultaneously divided in units of merging group using the horizontal component of the motion vector difference obtained by binarizer 314 with vertical
Measure 802x, y integer value.
Reconstructor 404 can draw video pictures are thin from the part for not including merging syntactic element of data flow 401
It is divided into block.Reconstructor 404 can be applicable to the horizontal component of the predetermined motion vector of all blocks of associated merging group
And vertical component, or refined by the horizontal component of the motion vector difference associated with the block of merging group and vertical component
The horizontal component and vertical component of the predetermined motion vector.
For the sake of complete, Figure 17 shows the encoder of suitable Figure 16 decoder.Figure 17 encoder includes structure
Make device 504, symbolism device 507 and entropy coder 513.Encoder includes constructor 504, and it is configured to swear by using motion
The motion compensated prediction of amount carrys out predictably encoded video 505 and the level of motion vector difference is set by predicted motion vector
The integer value 506 of component and vertical component is come predictably encoding motion vector to represent the prediction error of predicted motion vector；Symbol
Number change device 507, it is configured to binarization integer value and entered with obtaining the two of the horizontal component of motion vector difference and vertical component
Inhibition and generation 508, in the domain less than cutoff of horizontal component and vertical component, and block unitary code shape in cutoff respectively
In the prefix of formula and first interval in the combination of horizontal component and the suffix of the Exp- Golomb form of vertical component,
In the second interval in the domain for being more than and being equal to cutoff of horizontal component and vertical component, the binarization is respectively equal to water
The amount of dividing equally and vertical component block unitary code, and wherein cutoff is 2 and Exp- Golomb has order 1；And entropy code
Device 513, it is configured to, and for the horizontal component and vertical component of motion vector difference, utilizes context adaptive binary entropy
Coding will block unitary code and be encoded into data flow, and have lucky one relative to each binary digit position for blocking unitary code
Individual context, it is common for horizontal component and vertical component to motion vector difference, and is bypassed using constant equiprobability
Pattern-coding Exp- Golomb.It is further possible that realize that details can be changed directly from the description of the decoder on Figure 16
To Figure 17 encoder.
Although with regard to some aspects of context-descriptive of device, it is apparent that these aspects also illustrate that corresponding method
Description, wherein, square frame or device are corresponding with the feature of method and step or method and step.Similarly, in the upper of method and step
Aspect described in hereafter also illustrates that the description of the project or feature structure of corresponding square frame or corresponding device.Part is complete
Portion's method and step can be performed by (or use) hardware unit, similar such as microprocessor, programmable calculator or electronics electricity
Road.In certain embodiments, certain one in most important method and step or more persons can be performed by this device.
The encoded signal of the present invention can be stored on digital storage media or such as can wirelessly be passed in transmission medium
Defeated medium or wired transmissions medium such as interconnect transfers on network.
According to the requirement of some realizations, embodiments of the invention are realized in hardware or in software.Realize and number can be used
Word storage medium is carried out, such as floppy disk, DVD, Blu-ray disc, CD, ROM, PROM, EPROM, EEPROM or flash memory, is stored thereon with
Electronically readable takes control signal, and it cooperates (or can cooperate) with programmable computer system, so as to perform each method.Therefore
Digital storage media can be computer-readable medium.
Data medium with electronically readable control signal is included according to certain embodiments of the present invention, it can be with that can compile
Journey computer system cooperates, to perform one of methods described herein.
Under normal circumstances, embodiments of the invention can be realized as the computer program product with program code, when the meter
When calculation machine program product is run on computers, one of the program code operable the methods of being used for performing this.The program
Code is for example storable in machine-readable carrier.
Other embodiments are included based on performing one of methods described herein and being stored in machine-readable carrier
Calculation machine program.
In other words, therefore the embodiment of the inventive method is a kind of computer program with program code, for working as
The computer program is run on computers performs one of methods described herein.
Therefore, the another embodiment of the inventive method is a kind of data medium (or digital storage media or computer-readable
Medium) include recording thereon is used to perform the computer program of one of methods described herein.Data medium, numeral
Storage medium or recording medium are typically tangible and/or non-transformation.
Therefore, the another embodiment of the inventive method is used for the calculating for performing one of methods described herein for expression
The data flow or signal sequence of machine program.Data flow or signal sequence are for example configured to for example saturating via data communication connection
Cross internet transmission.
Another embodiment includes a kind of processing component, such as computer or programmable logic device, is configured to or is applicable
In one of execution methods described herein.
Further embodiment includes being provided with the computer program for performing one of methods described herein thereon
Computer.
Include a kind of device or system according to still another embodiment of the invention, it is configured to (such as electronically or optics
Ground) computer program of one of method will be described herein it is delivered to receiver for performing.Receiver may, for example, be
Computer, mobile device, memory devices etc..The device or system can be for example including for computer program to be transmitted
To the file server of receiver.
In certain embodiments, programmable logic device (such as field programmable gate array) can be used to perform this paper institutes
State part or all of function of method.In certain embodiments, field programmable gate array can be cooperated with microprocessor to hold
One of row methods described herein.Under normal circumstances, this methods of preferably performed by hardware unit.
Above-described embodiment is only used for illustrating the principle of the present invention.It should be understood that configuration as described herein and details are repaiied
Change and change and will be readily apparent to persons skilled in the art.Therefore, it is intended that the Patent right requirement only enclosed
Scope is limited, without being limited by the specific detail presented by the description and explanation of embodiment hereof.
Claims (32)
1. a kind of decoder being used for from data stream video, it will be moved using the binarization of horizontal component and vertical component
The horizontal component and vertical component of phasor difference are encoded in the data flow, in the horizontal component and vertical component less than cutoff
, and the prefix for blocking unitary code form in cutoff and the Exp- Golomb in horizontal component and vertical component respectively
In the first interval in the domain of the combination of the suffix of form, it is being more than and the horizontal component equal to cutoff and the domain of vertical component
In second interval, what the binarization was respectively equal to horizontal component and vertical component blocks unitary code, wherein cutoff be 2 and
Exp- Golomb has order 1, including：
Entropy decoder, it is configured to, for the horizontal component and vertical component of the motion vector difference, utilizes context-adaptive two
System entropy decoding from data flow draw it is described block unitary code, and each binary digit position for blocking unitary code have it is proper
What a context, it is common, and index brother human relations for horizontal component and vertical component to motion vector difference
Cloth code utilizes constant equiprobability bypass mode to obtain the binarization of the motion vector difference；
Inverse symbolism device, the binarization of the syntactic element of motion vector difference described in inverse binarization is configured to obtain
State the horizontal component of motion vector difference and the integer value of vertical component；
Reconstructor, the integer value for being configured to the horizontal component and vertical component of motion vector difference are described to rebuild
Video.
2. decoder according to claim 1, wherein the entropy decoder (409) is configured to utilize binary arithmetic decoding
Or binary system PIPE decodings draw from the data flow (401) and described block unitary code (806).
3. decoder according to claim 1 or 2, cut wherein the entropy decoder (409) is configured so that for described
The different contexts of two binary digit positions of disconnected unitary code (806).
4. decoder according to any one of claim 1 to 3, wherein the entropy decoder (409) is configured to pass through pin
To currently from the binary digit blocked unitary code (806) and drawn, from being selected for the binary digit currently drawn
The context-sensitive current probability State Transferring into the binary digit depending on currently drawing new probability shape
State updates to perform Probability State.
5. decoder according to any one of claim 1 to 4, wherein the entropy decoder (409) be configured to by with
Under type from described come to that currently will block the binary digit that unitary code (806) draws and carry out binarization arithmetic decoding：Quantify
The current probability interval width value in current probability section is represented to obtain probability interval index and be indexed by using probability interval
With depending on for the context-sensitive current probability state selected by the binary digit currently to be drawn
Probability State index be that table clause in table clause is indexed to carry out interval subdivision, to obtain current probability interval subdivision
For two partial sections.
6. decoder according to claim 5, wherein the entropy decoder (409) is configured so that for described current general
8 of rate interval width value represent and are configured to capture described 8 during the current probability interval width value is quantified
2 or 3 most significant bits represented.
7. the decoder according to claim 5 or 6, wherein the entropy decoder (409) is configured to work as from described
The bias state value of the inside in prior probability section selects from described two partial sections, update the probability interval width value and
The bias state value, and the partial section selected by utilization is inferred to the value for the binary digit currently to be drawn, and
The probability interval width value and bias state value of the renewal of continuity including the reading position from data flow (401) are carried out
Again standardize.
8. the decoder according to any one of claim 5 to 7, wherein the entropy decoder (409) is configured to, constant
Under equiprobability bypass mode, by current probability interval width value described in decile using obtain by the current probability interval subdivision as
The subdivision of two partial sections is so as to the binary digit progress binary arithmetic decoding from the Exp- Golomb.
9. decoder according to any one of claim 1 to 8, wherein entropy decoder (409) are configured to, for each fortune
Dynamic vector is poor, before the Exp- Golomb of the horizontal component and vertical component of each motion vector difference, from
The data flow show that the horizontal component of each motion vector difference and the described of the vertical component block unitary code.
10. decoder according to any one of claim 1 to 9, wherein the reconstructor be configured to from space and/or when
Between the horizontal component and vertical component of upper predicted motion vector be directed to described horizontal point of the motion vector to obtain
Measure with the fallout predictor of vertical component and by using described in the horizontal component and the vertical component refinement of the motion vector difference
Fallout predictor (826) rebuilds the horizontal component and vertical component of the motion vector.
11. decoder according to any one of claim 1 to 10, wherein the reconstructor is configured in a different manner
Come the horizontal component and vertical component of predicted motion vector and be directed to the horizontal component of motion vector to obtain and hang down
The sorted lists of the fallout predictor of straight component, list index is obtained and by using the institute of the motion vector difference from the data flow
The fallout predictor for stating the list that horizontal component and vertical component refine the list index sensing carrys out reconstructed motion vector
The horizontal component and vertical component.
12. the decoder according to claim 10 or 11, wherein the reconstructor is configured to lead to using motion compensated prediction
The horizontal component and vertical component using the motion vector are crossed to rebuild the video.
13. decoder according to claim 12, wherein the reconstructor is configured to lead to using the motion compensated prediction
The horizontal component for the spatial granularity application motion vector for crossing to be limited by the subdivision of the video pictures in block is divided with vertical
Measure to rebuild the video, wherein the reconstructor uses the merging syntactic element being present in the data flow so as to by described in
Block is grouped into merging group and applied in units of merging group by the motion vector difference of the inverse binarizer acquisition
The integer value of the horizontal component and vertical component.
14. decoder according to claim 13, wherein the reconstructor is configured to not include institute from the data flow
The part for stating merging syntactic element draws the subdivision of the video pictures in block.
15. the decoder according to claim 13 or 14, merge wherein the reconstructor is configured to use for associated
Group all blocks predetermined motion vector the horizontal component and vertical component, or by with the area for merging group
The horizontal component and vertical component of the associated motion vector difference of block refines the described of the predetermined motion vector
Horizontal component and vertical component.
16. a kind of encoder for being used to encode video into data flow, including：
Constructor, it is configured to predictably encode the video by motion compensated prediction, the motion compensated prediction utilizes fortune
Dynamic vector and by predict the motion vector and set motion vector difference horizontal component and vertical component integer value come in advance
Geodetic encodes the motion vector to represent the prediction error of the predicted motion vector；
Symbolism device, integer value described in binarization is configured to obtain the horizontal component of the motion vector difference and vertical
The binarization of component, in the horizontal component and vertical component less than cutoff, and blocking in cutoff respectively
The prefix of unitary code form and the domain of combination in the horizontal component and the suffix of the Exp- Golomb form of vertical component
First interval in, be more than and equal to the cutoff the horizontal component and vertical component domain second interval in,
What the binarization was respectively equal to the horizontal component and vertical component blocks unitary code, wherein the cutoff is 2 and institute
Stating Exp- Golomb has order 1；And
Entropy coder, it is configured to, it is adaptive using context for the horizontal component and vertical component of the motion vector difference
Answer binary system entropy code that the unitary code that blocks is encoded into the data flow, and each binary digit for blocking unitary code
Position has what a proper context, and it is common for the horizontal component and vertical component of the motion vector difference
, and Exp- Golomb coding utilizes constant equiprobability bypass mode.
17. encoder according to claim 16, wherein the entropy coder be configured to using binary arithmetic coding or
The unitary code that blocks is encoded into the data flow by binary system PIPE codings.
18. the encoder according to claim 16 or 17, blocked wherein the entropy coder is configured so that for described
The different contexts of two binary digit positions of unitary code.
19. the encoder according to any one of claim 16 to 18, wherein the entropy coder be configured to by for
Currently from it is described block unitary code coding binary digit, from for present encoding the binary digit selection it is described on
Hereafter associated current probability State Transferring performs into the new Probability State of the binary digit depending on currently drawing
Probability State updates.
20. the encoder according to any one of claim 16 to 19, wherein the entropy coder is configured to by following
Mode from the binary digit for blocking unitary code coding come to that currently will carry out binarization arithmetic coding：Quantization means are current
The current probability interval width value of probability interval with obtain probability interval index and by using probability interval index and depend on
With the Probability State of the context-sensitive current probability state selected for the binary digit currently to be drawn
Index and indexed for the table clause in table clause to carry out interval subdivision, to obtain the current probability interval subdivision as two
The subdivision of partial section.
21. encoder according to claim 20, wherein the entropy coder is configured so that for the current probability
8 of interval width value represent and are configured to capture 8 tables during the current probability interval width value is quantified
2 shown or 3 most significant bits.
22. the encoder according to claim 20 or 21, wherein the entropy coder is configured to what is currently encoded
The integer value of the binary digit selects from described two partial sections, utilizes selected partial section update probability area
Between width value and probability interval offset and to include to the data flow write position continuity the probability interval width value
Standardized again with the probability interval offset.
23. the encoder according to any one of claim 20 to 21, wherein the entropy coder is configured to pass through decile
The current probability interval width value is come subdivision of the current probability interval subdivision as two partial sections to coming using obtaining
Binary arithmetic coding is carried out from the binary digit of the Exp- Golomb.
24. the encoder according to any one of claim 16 to 23, wherein the entropy coder is configured to, for each
Motion vector difference, before the Exp- Golomb of the horizontal component and vertical component of each motion vector difference,
The data flow is encoded into by unitary code is blocked described in the horizontal component and vertical component of each motion vector difference.
25. the encoder according to any one of claim 16 to 24, wherein the constructor be configured to from space and/or
The horizontal component and vertical component of the motion vector are predicted on time to obtain the level of motion vector of being directed to
Component and the fallout predictor of vertical component simultaneously determine the horizontal component and vertical component of the motion vector difference so as to by described in
Fallout predictor is refined into the horizontal component and vertical component of motion vector.
26. the encoder according to any one of claim 16 to 25, wherein the constructor is configured to different sides
Formula is directed to described horizontal point of motion vector to predict the horizontal component and vertical component of the motion vector to obtain
The sorted lists of the fallout predictor of amount and vertical component, determine list index and the information for showing the list index is inserted into the number
According to flowing and determine the horizontal component and vertical component of the motion vector difference so as to the institute of list for pointing to list index
State the horizontal component and vertical component that fallout predictor is refined into motion vector.
27. the encoder according to any one of claim 16 to 26, wherein the constructor is configured to utilize the fortune
Dynamic compensation prediction passes through with the level of the spatial granularity application motion vector limited by the subdivision of the video pictures in block
Component and vertical component encode the video, wherein the constructor determines to merge syntactic element and is inserted into the data
Stream carries out binary system so as to which the block is grouped into merging group and applied in units of merging group by the inverse binarizer
The integer value of the horizontal component and vertical component of the motion vector difference changed.
28. encoder according to claim 27, wherein the constructor is configured to the video pictures in block
The subdivision be encoded into the data flow not include it is described merging syntactic element a part.
29. the encoder according to claim 27 or 28, merge wherein the constructor is configured to use for associated
Group all blocks predetermined motion vector the horizontal component and vertical component, or by with the area for merging group
The horizontal component and vertical component of the associated motion vector difference of block refines the described of the predetermined motion vector
Horizontal component and vertical component.
30. a kind of method being used for from data stream video, it will be moved using the binarization of horizontal component and vertical component
The horizontal component and vertical component of phasor difference are encoded to the data flow, in the horizontal component and vertical component less than cutoff
, and the prefix for blocking unitary code form in cutoff and the Exp- Golomb in horizontal component and vertical component respectively
In the first interval in the domain of the combination of the suffix of form, it is being more than and the horizontal component equal to cutoff and the domain of vertical component
In second interval, what the binarization was respectively equal to horizontal component and vertical component blocks unitary code, wherein cutoff be 2 and
Exp- Golomb has order 1,
Including：
For the horizontal component and vertical component of the motion vector difference, using context adaptive binary entropy decoding from data
Stream blocks unitary code described in drawing, and each binary digit position for blocking unitary code has what a proper context, its
It is common for the horizontal component and vertical component of motion vector difference, and the Exp- Golomb is general using constant grade
Rate bypass mode is to obtain the binarization of the motion vector difference；
The binarization of the syntactic element of motion vector difference described in inverse binarization is to obtain the water of the motion vector difference
The integer value of the amount of dividing equally and vertical component；
The integer value of the horizontal component and vertical component based on motion vector difference rebuilds the video.
31. a kind of encoder for being used to encode video into data flow, including：
The video is predictably encoded by motion compensated prediction, the motion compensated prediction is using motion vector and by pre-
Survey the motion vector and set the horizontal component of motion vector difference and the integer value of vertical component predictably to encode the fortune
Dynamic vector is to represent the prediction error of the predicted motion vector；
Integer value described in binarization to obtain the binarization of the horizontal component and vertical component of the motion vector difference,
In the horizontal component and vertical component less than cutoff, and it is in the prefix for blocking unitary code form of cutoff respectively
In the first interval in the domain of the combination in the horizontal component and the suffix of the Exp- Golomb form of vertical component, big
In the second interval in the domain of the horizontal component and vertical component of the cutoff, the binarization is respectively etc.
Unitary code is blocked in the horizontal component and vertical component, wherein the cutoff is 2 and the Exp- Golomb has
Sequentially 1；And
, will using context adaptive binary entropy code for the horizontal component and vertical component of the motion vector difference
The unitary code that blocks is encoded into the data flow, and each binary digit position for blocking unitary code have just what a
Context, it is common, and index brother human relations for the horizontal component and vertical component of the motion vector difference
Cloth code coding utilizes constant equiprobability bypass mode.
32. a kind of storage medium, is stored with computer program, the computer program, which has, to be used to carry out when running on computers
The program code of method according to any one of claim 30 to 31.
Applications Claiming Priority (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US201161497794P | 2011-06-16 | 2011-06-16 | |
| US61/497,794 | 2011-06-16 | ||
| US201161508506P | 2011-07-15 | 2011-07-15 | |
| US61/508,506 | 2011-07-15 | ||
| CN201280039896.3A CN103931194B (en) | 2011-06-16 | 2012-06-18 | The entropy code of motion vector difference |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201280039896.3A Division CN103931194B (en) | 2011-06-16 | 2012-06-18 | The entropy code of motion vector difference |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107529709A true CN107529709A (en) | 2018-01-02 |
| CN107529709B CN107529709B (en) | 2019-05-07 |
Family
ID=46319135
Family Applications (26)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710694424.6A Active CN107529708B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding and encoded video method and storage medium |
| CN201280039832.3A Active CN103733622B (en) | 2011-06-16 | 2012-06-18 | Context initialization in entropy code |
| CN201710965903.7A Active CN107613307B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, and storage medium |
| CN201710977462.2A Active CN107517384B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, and storage medium |
| CN201710694312.0A Active CN107801042B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video |
| CN201710694799.2A Active CN107529710B (en) | 2011-06-16 | 2012-06-18 | Decoder, method of decoding and encoding video |
| CN201280039896.3A Active CN103931194B (en) | 2011-06-16 | 2012-06-18 | The entropy code of motion vector difference |
| CN201280039922.2A Active CN103748886B (en) | 2011-06-16 | 2012-06-18 | Support the entropy code of pattern switching |
| CN201710694391.5A Active CN107347161B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710477936.7A Active CN107465928B (en) | 2011-06-16 | 2012-06-18 | Encoder for encoding video into data stream, decoder for decoding video from data stream, and encoding and decoding methods thereof |
| CN201710962467.8A Active CN107770559B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, medium, and storage method |
| CN201710478537.2A Active CN107333141B (en) | 2011-06-16 | 2012-06-18 | Decoder and decoding method, encoder and encoding method |
| CN201710491640.0A Active CN107333142B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, and decoding method and encoding method thereof |
| CN201710477937.1A Active CN107333140B (en) | 2011-06-16 | 2012-06-18 | Video coding and decoding method and coder and decoder |
| CN201710962381.5A Active CN107517383B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, and storage medium |
| CN201710694392.XA Active CN107529705B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710694423.1A Active CN107529707B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding and encoded video method |
| CN201710694354.4A Active CN107529704B (en) | 2011-06-16 | 2012-06-18 | Decoder, decoding and encoded video method |
| CN201710695327.9A Active CN107360423B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710694268.3A Active CN107801041B (en) | 2011-06-16 | 2012-06-18 | Decoder, decoding and encoded video method |
| CN201710477825.6A Active CN107465927B (en) | 2011-06-16 | 2012-06-18 | Decoder and decoding method for decoding video from data stream, and encoder and encoding method for encoding video into data stream |
| CN201710694425.0A Active CN107529709B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding and encoded video method and storage medium |
| CN201710694930.5A Active CN107360434B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710694292.7A Active CN107801025B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video |
| CN201710694395.3A Active CN107529706B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710477823.7A Active CN107465926B (en) | 2011-06-16 | 2012-06-18 | Context initialization-based decoder, encoding and decoding method, and computer medium |
Family Applications Before (21)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710694424.6A Active CN107529708B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding and encoded video method and storage medium |
| CN201280039832.3A Active CN103733622B (en) | 2011-06-16 | 2012-06-18 | Context initialization in entropy code |
| CN201710965903.7A Active CN107613307B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, and storage medium |
| CN201710977462.2A Active CN107517384B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, and storage medium |
| CN201710694312.0A Active CN107801042B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video |
| CN201710694799.2A Active CN107529710B (en) | 2011-06-16 | 2012-06-18 | Decoder, method of decoding and encoding video |
| CN201280039896.3A Active CN103931194B (en) | 2011-06-16 | 2012-06-18 | The entropy code of motion vector difference |
| CN201280039922.2A Active CN103748886B (en) | 2011-06-16 | 2012-06-18 | Support the entropy code of pattern switching |
| CN201710694391.5A Active CN107347161B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710477936.7A Active CN107465928B (en) | 2011-06-16 | 2012-06-18 | Encoder for encoding video into data stream, decoder for decoding video from data stream, and encoding and decoding methods thereof |
| CN201710962467.8A Active CN107770559B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, medium, and storage method |
| CN201710478537.2A Active CN107333141B (en) | 2011-06-16 | 2012-06-18 | Decoder and decoding method, encoder and encoding method |
| CN201710491640.0A Active CN107333142B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, and decoding method and encoding method thereof |
| CN201710477937.1A Active CN107333140B (en) | 2011-06-16 | 2012-06-18 | Video coding and decoding method and coder and decoder |
| CN201710962381.5A Active CN107517383B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding method, encoding method, and storage medium |
| CN201710694392.XA Active CN107529705B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710694423.1A Active CN107529707B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, decoding and encoded video method |
| CN201710694354.4A Active CN107529704B (en) | 2011-06-16 | 2012-06-18 | Decoder, decoding and encoded video method |
| CN201710695327.9A Active CN107360423B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710694268.3A Active CN107801041B (en) | 2011-06-16 | 2012-06-18 | Decoder, decoding and encoded video method |
| CN201710477825.6A Active CN107465927B (en) | 2011-06-16 | 2012-06-18 | Decoder and decoding method for decoding video from data stream, and encoder and encoding method for encoding video into data stream |
Family Applications After (4)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710694930.5A Active CN107360434B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710694292.7A Active CN107801025B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video |
| CN201710694395.3A Active CN107529706B (en) | 2011-06-16 | 2012-06-18 | Decoder, encoder, method of decoding and encoding video, and storage medium |
| CN201710477823.7A Active CN107465926B (en) | 2011-06-16 | 2012-06-18 | Context initialization-based decoder, encoding and decoding method, and computer medium |
Country Status (31)
| Country | Link |
|---|---|
| US (36) | US9918090B2 (en) |
| EP (9) | EP3703367A1 (en) |
| JP (11) | JP6059212B2 (en) |
| KR (13) | KR101730587B1 (en) |
| CN (26) | CN107529708B (en) |
| AP (3) | AP3686A (en) |
| AU (11) | AU2012268951B2 (en) |
| BR (4) | BR112013032333B1 (en) |
| CA (2) | CA2839569C (en) |
| CL (2) | CL2013003603A1 (en) |
| CO (2) | CO6852030A2 (en) |
| CY (1) | CY1124687T1 (en) |
| DK (5) | DK3471415T3 (en) |
| ES (5) | ES2961663T3 (en) |
| FI (2) | FI2721819T3 (en) |
| HK (17) | HK1197128A1 (en) |
| HR (1) | HRP20220272T1 (en) |
| HU (5) | HUE062605T2 (en) |
| IL (17) | IL312098A (en) |
| LT (4) | LT3930330T (en) |
| MX (7) | MX336735B (en) |
| MY (3) | MY170940A (en) |
| PH (15) | PH12018500331A1 (en) |
| PL (5) | PL3343781T3 (en) |
| PT (5) | PT3343781T (en) |
| RS (4) | RS63053B1 (en) |
| RU (7) | RU2658883C1 (en) |
| SI (4) | SI3930330T1 (en) |
| UA (2) | UA115186C2 (en) |
| WO (3) | WO2012172113A1 (en) |
| ZA (2) | ZA201400030B (en) |
Families Citing this family (228)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106454371B (en) | 2010-04-13 | 2020-03-20 | Ge视频压缩有限责任公司 | Decoder, array reconstruction method, encoder, encoding method, and storage medium |
| KR101584480B1 (en) | 2010-04-13 | 2016-01-14 | 지이 비디오 컴프레션, 엘엘씨 | Inter-plane prediction |
| KR102166520B1 (en) | 2010-04-13 | 2020-10-16 | 지이 비디오 컴프레션, 엘엘씨 | Sample region merging |
| ES2549734T3 (en) | 2010-04-13 | 2015-11-02 | Ge Video Compression, Llc | Video encoding using multi-tree image subdivisions |
| US9288089B2 (en) | 2010-04-30 | 2016-03-15 | Ecole Polytechnique Federale De Lausanne (Epfl) | Orthogonal differential vector signaling |
| US9246713B2 (en) | 2010-05-20 | 2016-01-26 | Kandou Labs, S.A. | Vector signaling with reduced receiver complexity |
| US9985634B2 (en) | 2010-05-20 | 2018-05-29 | Kandou Labs, S.A. | Data-driven voltage regulator |
| US9251873B1 (en) | 2010-05-20 | 2016-02-02 | Kandou Labs, S.A. | Methods and systems for pin-efficient memory controller interface using vector signaling codes for chip-to-chip communications |
| US9288082B1 (en) | 2010-05-20 | 2016-03-15 | Kandou Labs, S.A. | Circuits for efficient detection of vector signaling codes for chip-to-chip communication using sums of differences |
| US9077386B1 (en) | 2010-05-20 | 2015-07-07 | Kandou Labs, S.A. | Methods and systems for selection of unions of vector signaling codes for power and pin efficient chip-to-chip communication |
| US8593305B1 (en) | 2011-07-05 | 2013-11-26 | Kandou Labs, S.A. | Efficient processing and detection of balanced codes |
| WO2011151469A1 (en) | 2010-06-04 | 2011-12-08 | Ecole Polytechnique Federale De Lausanne | Error control coding for orthogonal differential vector signaling |
| US20120182388A1 (en) * | 2011-01-18 | 2012-07-19 | Samsung Electronics Co., Ltd. | Apparatus and method for processing depth image |
| BR112013032333B1 (en) | 2011-06-16 | 2022-07-26 | Ge Video Compression, Llc | AUXILIARY MODE SWITCHING FOR ENTROPY ENCODING |
| PL3309970T3 (en) | 2011-06-24 | 2023-09-25 | Sun Patent Trust | Image decoding method |
| DK2884749T3 (en) | 2011-06-28 | 2018-01-22 | Samsung Electronics Co Ltd | Apparatus for decoding video followed by arithmetic decoding and two-dimensional last significant coefficient signaling. |
| US11647197B2 (en) | 2011-06-30 | 2023-05-09 | Velos Media, Llc | Context initialization based on slice header flag and slice type |
| UA114674C2 (en) | 2011-07-15 | 2017-07-10 | ДЖ.І. ВІДІЕУ КЕМПРЕШН, ЛЛСі | CONTEXT INITIALIZATION IN ENTHROPIC CODING |
| BR112014001026B1 (en) | 2011-07-15 | 2022-05-03 | Ge Video Compression, Llc | Low-delay sample matrix encoding |
| EP2740272A4 (en) * | 2011-08-04 | 2015-04-08 | Mediatek Inc | Method and apparatus for reordered binarization of syntax elements in cabac |
| WO2013039692A1 (en) * | 2011-09-16 | 2013-03-21 | Mediatek Singapore Pte. Ltd. | Method and apparatus for prediction mode and partition mode syntax coding for coding units in hevc |
| WO2013050612A1 (en) * | 2011-10-06 | 2013-04-11 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Entropy coding buffer arrangement |
| US20130101047A1 (en) * | 2011-10-19 | 2013-04-25 | Sony Corporation | Context reduction of significance map coding of 4x4 and 8x8 transform coefficient in hm4.0 |
| US9871537B2 (en) * | 2011-10-27 | 2018-01-16 | Qualcomm Incorporated | Mapping states in binary arithmetic coder for video coding |
| US9484952B2 (en) | 2011-11-03 | 2016-11-01 | Qualcomm Incorporated | Context state and probability initialization for context adaptive entropy coding |
| IN2014CN03598A (en) * | 2011-11-04 | 2015-07-31 | Sharp Kk | |
| KR20130050403A (en) * | 2011-11-07 | 2013-05-16 | 오수미 | Method for generating rrconstructed block in inter prediction mode |
| US9503717B2 (en) | 2012-01-09 | 2016-11-22 | Texas Instruments Incorporated | Context adaptive binary arithmetic coding (CABAC) with scalable throughput and coding efficiency |
| WO2013106987A1 (en) * | 2012-01-16 | 2013-07-25 | Mediatek Singapore Pte. Ltd. | Methods and apparatuses of bypass coding and reducing contexts for some syntax elements |
| US20130188698A1 (en) * | 2012-01-19 | 2013-07-25 | Qualcomm Incorporated | Coefficient level coding |
| US9866829B2 (en) * | 2012-01-22 | 2018-01-09 | Qualcomm Incorporated | Coding of syntax elements that correspond to coefficients of a coefficient block in video coding |
| JP6156723B2 (en) * | 2012-03-08 | 2017-07-05 | サン パテント トラスト | Image encoding method, image decoding method, image encoding device, and image decoding device |
| US11039138B1 (en) | 2012-03-08 | 2021-06-15 | Google Llc | Adaptive coding of prediction modes using probability distributions |
| US9237344B2 (en) * | 2012-03-22 | 2016-01-12 | Qualcomm Incorporated | Deriving context for last position coding for video coding |
| US9584802B2 (en) | 2012-04-13 | 2017-02-28 | Texas Instruments Incorporated | Reducing context coded and bypass coded bins to improve context adaptive binary arithmetic coding (CABAC) throughput |
| US9621921B2 (en) | 2012-04-16 | 2017-04-11 | Qualcomm Incorporated | Coefficient groups and coefficient coding for coefficient scans |
| WO2013175736A1 (en) * | 2012-05-25 | 2013-11-28 | パナソニック株式会社 | Video encoding method, video encoding device, video decoding method, video decoding device, and video encoding/decoding device |
| EP2858354B1 (en) | 2012-05-25 | 2022-12-21 | Sun Patent Trust | Video image coding method, video image decoding method, video image coding device, video image decoding device, and video image coding-decoding device |
| WO2013175698A1 (en) | 2012-05-25 | 2013-11-28 | パナソニック株式会社 | Image encoding method, image encoding device, image decoding method, image decoding device, and image encoding/decoding device |
| EP2858357B1 (en) | 2012-06-04 | 2022-07-06 | Sun Patent Trust | Video image encoding method, video image decoding method |
| DE102012224060A1 (en) * | 2012-12-20 | 2014-06-26 | Robert Bosch Gmbh | Method for data transmission for a battery management system |
| CN104995612B (en) | 2013-01-17 | 2020-01-03 | 康杜实验室公司 | Inter-chip communication method and system with low synchronous switching noise |
| CN105122758B (en) | 2013-02-11 | 2018-07-10 | 康杜实验室公司 | High bandwidth interchip communication interface method and system |
| CN104053007B (en) * | 2013-03-15 | 2019-05-07 | 乐金电子(中国)研究开发中心有限公司 | Depth image intra-frame encoding mode index value configuration method and device |
| GB2513111A (en) | 2013-04-08 | 2014-10-22 | Sony Corp | Data encoding and decoding |
| KR102241045B1 (en) | 2013-04-16 | 2021-04-19 | 칸도우 랩스 에스에이 | Methods and systems for high bandwidth communications interface |
| WO2014210074A1 (en) | 2013-06-25 | 2014-12-31 | Kandou Labs SA | Vector signaling with reduced receiver complexity |
| CN105359508A (en) | 2013-06-25 | 2016-02-24 | 努梅利有限公司 | Multi-level spatial-temporal resolution increase of video |
| US10021419B2 (en) * | 2013-07-12 | 2018-07-10 | Qualcomm Incorported | Rice parameter initialization for coefficient level coding in video coding process |
| US20150098504A1 (en) * | 2013-10-09 | 2015-04-09 | Qualcomm Incorporated | Block vector coding for intra block copying |
| US10270590B2 (en) | 2013-12-16 | 2019-04-23 | Mcafee, Llc | Process efficient preprocessing for any encryption standard |
| GB2521828A (en) * | 2013-12-23 | 2015-07-08 | Sony Corp | Data encoding and decoding |
| WO2015100522A1 (en) * | 2013-12-30 | 2015-07-09 | Mediatek Singapore Pte. Ltd. | Methods for inter-component residual prediction |
| US9806761B1 (en) | 2014-01-31 | 2017-10-31 | Kandou Labs, S.A. | Methods and systems for reduction of nearest-neighbor crosstalk |
| CN110266615B (en) | 2014-02-02 | 2022-04-29 | 康杜实验室公司 | Low-power inter-chip communication method and device with low ISI ratio |
| US10885104B2 (en) * | 2014-02-27 | 2021-01-05 | Dropbox, Inc. | Systems and methods for selecting content items to store and present locally on a user device |
| US9787799B2 (en) | 2014-02-27 | 2017-10-10 | Dropbox, Inc. | Systems and methods for managing content items having multiple resolutions |
| US9363114B2 (en) | 2014-02-28 | 2016-06-07 | Kandou Labs, S.A. | Clock-embedded vector signaling codes |
| US9509437B2 (en) | 2014-05-13 | 2016-11-29 | Kandou Labs, S.A. | Vector signaling code with improved noise margin |
| US20150334425A1 (en) * | 2014-05-14 | 2015-11-19 | Blackberry Limited | Adaptive context initialization |
| US9148087B1 (en) | 2014-05-16 | 2015-09-29 | Kandou Labs, S.A. | Symmetric is linear equalization circuit with increased gain |
| US10848786B2 (en) * | 2014-05-28 | 2020-11-24 | Arris Enterprises Llc | Acceleration of context adaptive binary arithmetic coding (CABAC) in video CODECS |
| US9337862B2 (en) | 2014-06-09 | 2016-05-10 | Tidal Systems, Inc. | VLSI efficient Huffman encoding apparatus and method |
| US9930341B2 (en) * | 2014-06-20 | 2018-03-27 | Qualcomm Incorporated | Block vector coding for intra block copying |
| US9852806B2 (en) | 2014-06-20 | 2017-12-26 | Kandou Labs, S.A. | System for generating a test pattern to detect and isolate stuck faults for an interface using transition coding |
| US9112550B1 (en) | 2014-06-25 | 2015-08-18 | Kandou Labs, SA | Multilevel driver for high speed chip-to-chip communications |
| US9854253B2 (en) * | 2014-06-30 | 2017-12-26 | Qualcomm Incorporated | Method for motion vector difference (MVD) and intra block copy vector difference (BVD) coding of screen content video data |
| CN106797352B (en) | 2014-07-10 | 2020-04-07 | 康杜实验室公司 | High signal-to-noise characteristic vector signaling code |
| US9432082B2 (en) | 2014-07-17 | 2016-08-30 | Kandou Labs, S.A. | Bus reversable orthogonal differential vector signaling codes |
| KR101943048B1 (en) | 2014-07-21 | 2019-01-28 | 칸도우 랩스 에스에이 | Multidrop data transfer |
| EP3175592B1 (en) | 2014-08-01 | 2021-12-29 | Kandou Labs S.A. | Orthogonal differential vector signaling codes with embedded clock |
| US10283091B2 (en) * | 2014-10-13 | 2019-05-07 | Microsoft Technology Licensing, Llc | Buffer optimization |
| CN105578180B (en) * | 2014-10-16 | 2019-01-15 | 联想(北京)有限公司 | A kind of coding method and device |
| US9674014B2 (en) | 2014-10-22 | 2017-06-06 | Kandou Labs, S.A. | Method and apparatus for high speed chip-to-chip communications |
| CN104392725A (en) * | 2014-12-02 | 2015-03-04 | 中科开元信息技术(北京)有限公司 | Method and device for hybrid coding/decoding of multi-channel lossless audios |
| US10097833B2 (en) * | 2014-12-26 | 2018-10-09 | Intel Corporation | Method and system of entropy coding using look-up table based probability updating for video coding |
| WO2016119726A1 (en) * | 2015-01-30 | 2016-08-04 | Mediatek Inc. | Method and apparatus for entropy coding of source samples with large alphabet |
| US9942551B2 (en) * | 2015-01-30 | 2018-04-10 | Qualcomm Incorporated | Palette index grouping for video coding |
| WO2016124710A1 (en) * | 2015-02-05 | 2016-08-11 | Fraunhofer Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Multi-view video codec supporting residual prediction |
| US9930378B2 (en) * | 2015-02-11 | 2018-03-27 | Qualcomm Incorporated | Signaling of operation points for carriage of HEVC extensions |
| US10958927B2 (en) | 2015-03-27 | 2021-03-23 | Qualcomm Incorporated | Motion information derivation mode determination in video coding |
| SG11201708026TA (en) * | 2015-04-08 | 2017-10-30 | Hfi Innovation Inc | Methods of palette mode coding in video coding |
| EP3292687A4 (en) * | 2015-05-06 | 2019-07-03 | NGCodec Inc. | Intra prediction processor with reduced cost block partitioning and refined intra mode selection |
| US10225555B2 (en) * | 2015-05-19 | 2019-03-05 | Mediatek Inc. | Method and apparatus for multi-table based context adaptive binary arithmetic coding |
| WO2016210445A1 (en) | 2015-06-26 | 2016-12-29 | Kandou Labs, S.A. | High speed communications system |
| KR20240000649A (en) * | 2015-08-24 | 2024-01-02 | 엘지전자 주식회사 | Inter prediction mode-based image processing method and apparatus therefor |
| WO2017041271A1 (en) * | 2015-09-10 | 2017-03-16 | Mediatek Singapore Pte. Ltd. | Efficient context modeling for coding a block of data |
| CN114697679A (en) * | 2015-09-11 | 2022-07-01 | 株式会社Kt | Image decoding method, image encoding method, and apparatus including bit stream |
| US10158874B2 (en) * | 2015-09-30 | 2018-12-18 | Apple Inc. | Parallel bypass and regular bin coding |
| US10198355B2 (en) | 2015-10-29 | 2019-02-05 | Dropbox, Inc. | Proving a dynamic digital content cache |
| US10440399B2 (en) * | 2015-11-13 | 2019-10-08 | Qualcomm Incorporated | Coding sign information of video data |
| US10055372B2 (en) | 2015-11-25 | 2018-08-21 | Kandou Labs, S.A. | Orthogonal differential vector signaling codes with embedded clock |
| US10003315B2 (en) | 2016-01-25 | 2018-06-19 | Kandou Labs S.A. | Voltage sampler driver with enhanced high-frequency gain |
| FR3047379A1 (en) * | 2016-01-29 | 2017-08-04 | Orange | METHOD FOR ENCODING AND DECODING DATA, DEVICE FOR ENCODING AND DECODING DATA AND CORRESPONDING COMPUTER PROGRAMS |
| US10791097B2 (en) | 2016-04-14 | 2020-09-29 | Sophos Limited | Portable encryption format |
| US10573324B2 (en) | 2016-02-24 | 2020-02-25 | Dolby International Ab | Method and system for bit reservoir control in case of varying metadata |
| CN109314518B (en) | 2016-04-22 | 2022-07-29 | 康杜实验室公司 | High performance phase locked loop |
| US10003454B2 (en) | 2016-04-22 | 2018-06-19 | Kandou Labs, S.A. | Sampler with low input kickback |
| US10056903B2 (en) | 2016-04-28 | 2018-08-21 | Kandou Labs, S.A. | Low power multilevel driver |
| US10153591B2 (en) | 2016-04-28 | 2018-12-11 | Kandou Labs, S.A. | Skew-resistant multi-wire channel |
| US10333741B2 (en) | 2016-04-28 | 2019-06-25 | Kandou Labs, S.A. | Vector signaling codes for densely-routed wire groups |
| US10708164B2 (en) * | 2016-05-03 | 2020-07-07 | Qualcomm Incorporated | Binarizing secondary transform index |
| US10142652B2 (en) * | 2016-05-05 | 2018-11-27 | Google Llc | Entropy coding motion vector residuals obtained using reference motion vectors |
| FR3054764B1 (en) * | 2016-07-28 | 2018-07-27 | B<>Com | METHOD OF DECODING AN IMAGE, ENCODING METHOD, DEVICES, TERMINAL EQUIPMENT AND COMPUTER PROGRAMS |
| WO2018030293A1 (en) * | 2016-08-10 | 2018-02-15 | パナソニック インテレクチュアル プロパティ コーポレーション オブ アメリカ | Encoding device, decoding device, encoding method and decoding method |
| US9906358B1 (en) | 2016-08-31 | 2018-02-27 | Kandou Labs, S.A. | Lock detector for phase lock loop |
| US10411922B2 (en) | 2016-09-16 | 2019-09-10 | Kandou Labs, S.A. | Data-driven phase detector element for phase locked loops |
| US10419781B2 (en) * | 2016-09-20 | 2019-09-17 | Qualcomm Incorporated | Storing and retrieving high bit depth image data |
| US10264264B2 (en) * | 2016-09-24 | 2019-04-16 | Apple Inc. | Multi-bin decoding systems and methods |
| US10462462B2 (en) * | 2016-09-29 | 2019-10-29 | Qualcomm Incorporated | Motion vector difference coding technique for video coding |
| US11356693B2 (en) * | 2016-09-29 | 2022-06-07 | Qualcomm Incorporated | Motion vector coding for video coding |
| CN106231318B (en) * | 2016-09-30 | 2020-04-28 | 浙江宇视科技有限公司 | Video scrambling method and device based on quantized transform coefficients |
| AU2017342688A1 (en) * | 2016-10-11 | 2019-05-02 | Genomsys Sa | Method and system for selective access of stored or transmitted bioinformatics data |
| US10200188B2 (en) | 2016-10-21 | 2019-02-05 | Kandou Labs, S.A. | Quadrature and duty cycle error correction in matrix phase lock loop |
| US10200218B2 (en) | 2016-10-24 | 2019-02-05 | Kandou Labs, S.A. | Multi-stage sampler with increased gain |
| US10372665B2 (en) | 2016-10-24 | 2019-08-06 | Kandou Labs, S.A. | Multiphase data receiver with distributed DFE |
| MX2019007219A (en) * | 2016-12-22 | 2019-09-05 | Kt Corp | Video signal processing method and device. |
| WO2018124855A1 (en) * | 2017-01-02 | 2018-07-05 | 한양대학교 산학협력단 | Method and apparatus for decoding image using inter-picture prediction |
| WO2018128222A1 (en) * | 2017-01-03 | 2018-07-12 | 엘지전자 주식회사 | Method and apparatus for image decoding in image coding system |
| US10757412B2 (en) * | 2017-01-03 | 2020-08-25 | Avago Technologies International Sales Pte. Limited | Architecture flexible binary arithmetic coding system |
| KR20180087771A (en) * | 2017-01-25 | 2018-08-02 | 삼성전자주식회사 | Entropy encoder and video encoder including the same |
| US10595033B2 (en) | 2017-02-15 | 2020-03-17 | Sony Corporation | Variable length coding of header data for image compression |
| US10554988B2 (en) * | 2017-03-22 | 2020-02-04 | Qualcomm Incorporated | Binary arithmetic coding with parameterized probability estimation finite state machines |
| KR102450863B1 (en) * | 2017-03-22 | 2022-10-05 | 에스케이텔레콤 주식회사 | Method and Apparatus for Encoding and Decoding Motion Vector |
| MX2019012293A (en) | 2017-04-13 | 2020-01-20 | Lg Electronics Inc | Method and device for entropy encoding, decoding video signal. |
| CN110741562B (en) | 2017-04-14 | 2022-11-04 | 康杜实验室公司 | Pipelined forward error correction for vector signaling code channels |
| US10116468B1 (en) | 2017-06-28 | 2018-10-30 | Kandou Labs, S.A. | Low power chip-to-chip bidirectional communications |
| US10686583B2 (en) | 2017-07-04 | 2020-06-16 | Kandou Labs, S.A. | Method for measuring and correcting multi-wire skew |
| US10693587B2 (en) | 2017-07-10 | 2020-06-23 | Kandou Labs, S.A. | Multi-wire permuted forward error correction |
| US11477492B2 (en) * | 2017-08-04 | 2022-10-18 | Google Inc. | Adaptation for entropy coding of blocks of image data |
| US10203226B1 (en) | 2017-08-11 | 2019-02-12 | Kandou Labs, S.A. | Phase interpolation circuit |
| US10791341B2 (en) * | 2017-10-10 | 2020-09-29 | Qualcomm Incorporated | Binary arithmetic coding with progressive modification of adaptation parameters |
| US10484695B2 (en) | 2017-10-23 | 2019-11-19 | Google Llc | Refined entropy coding for level maps |
| US11039143B2 (en) * | 2017-11-20 | 2021-06-15 | Qualcomm Incorporated | Memory reduction for context initialization with temporal prediction |
| EP3490253A1 (en) * | 2017-11-23 | 2019-05-29 | Thomson Licensing | Encoding and decoding methods and corresponding devices |
| US11297339B2 (en) * | 2017-12-06 | 2022-04-05 | V-Nova International Limited | Methods and apparatuses for hierarchically encoding and decoding a bytestream |
| US10326623B1 (en) | 2017-12-08 | 2019-06-18 | Kandou Labs, S.A. | Methods and systems for providing multi-stage distributed decision feedback equalization |
| US10609421B2 (en) * | 2018-07-12 | 2020-03-31 | Google Llc | Context derivation for coefficient coding |
| US10694205B2 (en) * | 2017-12-18 | 2020-06-23 | Google Llc | Entropy coding of motion vectors using categories of transform blocks |
| US10602178B1 (en) * | 2017-12-21 | 2020-03-24 | Mozilla Corporation | Systems and methods for frame context selection |
| EP3503557A1 (en) * | 2017-12-22 | 2019-06-26 | Thomson Licensing | Method and apparatus for video encoding and decoding based on context switching |
| US10554380B2 (en) | 2018-01-26 | 2020-02-04 | Kandou Labs, S.A. | Dynamically weighted exclusive or gate having weighted output segments for phase detection and phase interpolation |
| KR20200112945A (en) * | 2018-01-29 | 2020-10-05 | 인터디지털 브이씨 홀딩스 인코포레이티드 | Encoding and decoding with refinement of reconstructed pictures |
| JP7187572B2 (en) * | 2018-03-29 | 2022-12-12 | フラウンホーファー－ゲゼルシャフト・ツール・フェルデルング・デル・アンゲヴァンテン・フォルシュング・アインゲトラーゲネル・フェライン | Enhancement concept of parallel coding function |
| JP7181941B2 (en) * | 2018-03-29 | 2022-12-01 | フラウンホーファー－ゲゼルシャフト・ツール・フェルデルング・デル・アンゲヴァンテン・フォルシュング・アインゲトラーゲネル・フェライン | video decoder, video encoder, method of decoding video content, method of encoding video content, computer program, and video bitstream |
| US10986354B2 (en) * | 2018-04-16 | 2021-04-20 | Panasonic Intellectual Property Corporation Of America | Encoder, decoder, encoding method, and decoding method |
| US10645381B2 (en) | 2018-04-30 | 2020-05-05 | Google Llc | Intra-prediction for smooth blocks in image/video |
| US10630990B1 (en) * | 2018-05-01 | 2020-04-21 | Amazon Technologies, Inc. | Encoder output responsive to quality metric information |
| CN108810553B (en) * | 2018-05-02 | 2021-01-12 | 上海大学 | Mobile node monitoring data sequence compression method based on sparse processing |
| TWI675441B (en) * | 2018-05-14 | 2019-10-21 | 欣興電子股份有限公司 | Package carrier structure and manufacturing method thereof |
| KR102559966B1 (en) * | 2018-05-25 | 2023-07-26 | 라인플러스 주식회사 | Method and system for transmitting and reproducing video of dynamic bit rate using a plurality of channels |
| EP4283998A1 (en) | 2018-05-30 | 2023-11-29 | Huawei Technologies Co., Ltd. | Multi-type tree depth extension for picture boundary handling |
| US10986340B2 (en) * | 2018-06-01 | 2021-04-20 | Qualcomm Incorporated | Coding adaptive multiple transform information for video coding |
| US10939115B2 (en) * | 2018-06-19 | 2021-03-02 | Qualcomm Incorporated | Efficient weighted probability estimation for binary arithmetic coding |
| CN108848380B (en) * | 2018-06-20 | 2021-11-30 | 腾讯科技（深圳）有限公司 | Video encoding and decoding method, device, computer device and storage medium |
| WO2020008329A1 (en) | 2018-07-01 | 2020-01-09 | Beijing Bytedance Network Technology Co., Ltd. | Spatial motion compression |
| CN112369025A (en) * | 2018-07-02 | 2021-02-12 | 交互数字Vc控股公司 | Context-based binary arithmetic coding and decoding |
| US11051025B2 (en) * | 2018-07-13 | 2021-06-29 | Tencent America LLC | Method and apparatus for video coding |
| CN108989825B (en) * | 2018-07-18 | 2021-05-07 | 北京奇艺世纪科技有限公司 | Arithmetic coding method and device and electronic equipment |
| US11218737B2 (en) | 2018-07-23 | 2022-01-04 | Google Llc | Asymmetric probability model update and entropy coding precision |
| US10743029B2 (en) * | 2018-07-30 | 2020-08-11 | Tencent America LLC | Constraints on coding unit partition |
| CN110809155B (en) | 2018-08-04 | 2023-01-31 | 北京字节跳动网络技术有限公司 | Video processing method, device and computer readable medium |
| US11336918B2 (en) | 2018-09-05 | 2022-05-17 | Qualcomm Incorporated | Regular coded bin reduction for coefficient coding |
| CN112040247B (en) * | 2018-09-10 | 2021-09-21 | 华为技术有限公司 | Video decoding method, video decoder, and computer-readable storage medium |
| CN114125467A (en) * | 2018-09-13 | 2022-03-01 | 华为技术有限公司 | Decoding method and device for predicting motion information |
| US11006150B2 (en) * | 2018-09-24 | 2021-05-11 | Tencent America LLC | Method and apparatus for video coding |
| KR102657902B1 (en) * | 2018-10-05 | 2024-04-17 | 로즈데일 다이나믹스 엘엘씨 | Transform coefficient coding method and device therefor |
| CN111010569B (en) | 2018-10-06 | 2023-02-28 | 北京字节跳动网络技术有限公司 | Improvement of temporal gradient calculation in BIO |
| US20200143226A1 (en) * | 2018-11-05 | 2020-05-07 | Samsung Electronics Co., Ltd. | Lossy compression of neural network activation maps |
| US10983677B2 (en) | 2018-11-16 | 2021-04-20 | Dropbox, Inc. | Prefetching digital thumbnails from remote servers to client devices based on a dynamic determination of file display criteria |
| US10841595B2 (en) * | 2018-11-27 | 2020-11-17 | Semiconductor Components Industries, Llc | Methods and apparatus for transform coefficient encoding and decoding |
| US10917636B2 (en) * | 2018-12-03 | 2021-02-09 | Tencent America LLC | Method and apparatus for video coding |
| US10904548B2 (en) * | 2018-12-06 | 2021-01-26 | Qualcomm Incorporated | Coefficient processing for video encoding and decoding |
| US11323748B2 (en) | 2018-12-19 | 2022-05-03 | Qualcomm Incorporated | Tree-based transform unit (TU) partition for video coding |
| CN115442616A (en) * | 2018-12-21 | 2022-12-06 | 华为技术有限公司 | Video decoding method and device |
| CN115209144B (en) * | 2018-12-29 | 2024-01-02 | 华为技术有限公司 | Encoder, decoder and corresponding methods using compressed MV storage |
| US11252431B2 (en) | 2019-01-02 | 2022-02-15 | Telefonaktiebolaget Lm Ericsson (Publ) | Side motion refinement in video encoding/decoding systems |
| US11032560B2 (en) * | 2019-01-17 | 2021-06-08 | Tencent America LLC | Method and apparatus for video coding without updating the HMVP table |
| CN111475424B (en) * | 2019-01-23 | 2023-07-28 | 伊姆西Ip控股有限责任公司 | Method, apparatus, and computer readable storage medium for managing a storage system |
| US11134273B2 (en) | 2019-02-14 | 2021-09-28 | Qualcomm Incorporated | Regular coded bin reduction for coefficient coding |
| CN109946229B (en) * | 2019-02-25 | 2024-07-16 | 南京文易特电子科技有限公司 | Intelligent digital double-pull-wire detection system and detection method for cigarette strip package |
| US10652581B1 (en) * | 2019-02-27 | 2020-05-12 | Google Llc | Entropy coding in image and video compression using machine learning |
| US10939107B2 (en) * | 2019-03-01 | 2021-03-02 | Sony Corporation | Embedded codec circuitry for sub-block based allocation of refinement bits |
| WO2020180153A1 (en) * | 2019-03-06 | 2020-09-10 | 엘지전자 주식회사 | Method and apparatus for processing video signal for inter prediction |
| US11202100B2 (en) * | 2019-03-11 | 2021-12-14 | Qualcomm Incorporated | Coefficient coding for transform skip mode |
| EP3939285A1 (en) * | 2019-03-12 | 2022-01-19 | InterDigital VC Holdings, Inc. | Flexible allocation of regular bins in residual coding for video coding |
| US11178399B2 (en) * | 2019-03-12 | 2021-11-16 | Qualcomm Incorporated | Probability initialization for video coding |
| WO2020189978A1 (en) * | 2019-03-15 | 2020-09-24 | 삼성전자 주식회사 | Method and device for decoding video, and method and device for encoding video |
| KR20220007853A (en) * | 2019-03-18 | 2022-01-19 | 프라운호퍼 게젤샤프트 쭈르 푀르데룽 데어 안겐반텐 포르슝 에. 베. | Method and apparatus for compressing parameters of a neural network |
| CN109947398B (en) * | 2019-03-25 | 2020-12-25 | 武汉轻工大学 | Triple integral solving method and device, terminal equipment and readable storage medium |
| CN113678462B (en) | 2019-04-15 | 2023-01-10 | 北京字节跳动网络技术有限公司 | Clipping parameter derivation in adaptive loop filters |
| KR20210135337A (en) | 2019-05-14 | 2021-11-12 | 엘지전자 주식회사 | Video or image coding based on adaptive loop filter |
| CN112135149B (en) * | 2019-06-24 | 2023-07-18 | 华为技术有限公司 | Entropy encoding/decoding method and device of syntax element and codec |
| US11184642B2 (en) | 2019-08-02 | 2021-11-23 | Mediatek Inc. | Methods and apparatus for simplification of coding residual blocks |
| KR20210018140A (en) * | 2019-08-06 | 2021-02-17 | 현대자동차주식회사 | Entropy-coding for Video Encoding and Decoding |
| WO2021040492A1 (en) * | 2019-08-31 | 2021-03-04 | 엘지전자 주식회사 | Method and apparatus for coding transform coefficient in video/image coding system |
| CN114731446A (en) * | 2019-09-24 | 2022-07-08 | 弗劳恩霍夫应用研究促进协会 | Coding concept for sequences of information values |
| WO2021058655A1 (en) * | 2019-09-24 | 2021-04-01 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Arithmetic encoders and decoders, video encoders and decoders, methods for encoding or decoding symbols, and methods for encoding or decoding video content |
| KR102626260B1 (en) * | 2019-10-01 | 2024-01-17 | 베이징 다지아 인터넷 인포메이션 테크놀로지 컴퍼니 리미티드 | Residual and coefficient coding method and device |
| US20230007260A1 (en) * | 2019-11-08 | 2023-01-05 | Google Llc | Probability Estimation for Video Coding |
| CN112866181B (en) * | 2019-11-28 | 2023-05-26 | 上海商汤智能科技有限公司 | Data decoding device, accelerator, and system on chip |
| US11468601B2 (en) | 2020-01-13 | 2022-10-11 | Samsung Display Co., Ltd. | Systems and methods for scalable throughput entropy coder |
| US11822902B2 (en) * | 2020-01-21 | 2023-11-21 | California Institute Of Technology | Systems and methods for performing lossless source coding |
| EP4078957A4 (en) * | 2020-02-12 | 2024-01-03 | Google LLC | Multi-context entropy coding for compression of graphs |
| GB2593691B (en) * | 2020-03-30 | 2022-08-24 | Imagination Tech Ltd | Efficient encoding methods |
| WO2022008311A1 (en) * | 2020-07-10 | 2022-01-13 | Koninklijke Philips N.V. | Genomic information compression by configurable machine learning-based arithmetic coding |
| US11362681B2 (en) | 2020-08-21 | 2022-06-14 | Seagate Technology Llc | One-shot state transition probability encoder and decoder |
| CN112260699B (en) * | 2020-09-15 | 2024-06-28 | 深圳拓邦股份有限公司 | Attribute communication coding and decoding method, coding device, decoding device and system |
| FR3114717B1 (en) * | 2020-09-30 | 2023-10-13 | Fond B Com | Method and electronic device for decoding a data stream, computer program and associated data stream |
| US11722672B2 (en) | 2020-10-06 | 2023-08-08 | Qualcomm Incorporated | Adaptively deriving rice parameter values for high bit-depth video coding |
| CN114501036A (en) * | 2020-11-13 | 2022-05-13 | 联发科技股份有限公司 | Entropy decoding device and related entropy decoding method |
| CN114727109B (en) * | 2021-01-05 | 2023-03-24 | 腾讯科技（深圳）有限公司 | Multimedia quantization processing method and device and coding and decoding equipment |
| CN115086657A (en) * | 2021-03-14 | 2022-09-20 | 腾讯科技（深圳）有限公司 | Video encoding and decoding method and device, computer readable medium and electronic equipment |
| US11356197B1 (en) | 2021-03-19 | 2022-06-07 | Kandou Labs SA | Error-tolerant forward error correction ordered set message decoder |
| CN113079377B (en) * | 2021-04-01 | 2022-09-30 | 中国科学技术大学 | Training method for depth image/video compression network |
| WO2022217245A1 (en) * | 2021-04-07 | 2022-10-13 | Innopeak Technology, Inc. | Remaining level binarization for video coding |
| CN113489979A (en) * | 2021-05-28 | 2021-10-08 | 杭州博雅鸿图视频技术有限公司 | Entropy coding method, entropy coding device, electronic equipment and storage medium |
| CN113422964B (en) * | 2021-06-04 | 2024-07-30 | 绍兴市北大信息技术科创中心 | String length parameter coding and decoding method and device |
| KR20240044497A (en) * | 2021-08-17 | 2024-04-04 | 베이징 다지아 인터넷 인포메이션 테크놀로지 컴퍼니 리미티드 | Sign prediction for block-based video coding |
| US11871038B2 (en) * | 2021-09-29 | 2024-01-09 | Tencent America LLC | Techniques for constraint flag signaling for range extension with residual rice coding extension |
| EP4449722A1 (en) * | 2021-12-16 | 2024-10-23 | Beijing Dajia Internet Information Technology Co., Ltd. | Sign prediction for block-based video coding |
| US20230247206A1 (en) * | 2022-02-03 | 2023-08-03 | Tencent America LLC | Methods, devices, and storage medium for multi-symbol arithmetic coding |
| US20230291935A1 (en) * | 2022-03-11 | 2023-09-14 | Tencent America LLC | Systems and methods for division-free probability regularization for arithmetic coding |
| US20230308651A1 (en) * | 2022-03-22 | 2023-09-28 | Tencent America LLC | Systems and methods for regularization-free multi-hypothesis arithmetic coding |
| WO2023195643A1 (en) * | 2022-04-05 | 2023-10-12 | 삼성전자 주식회사 | Device for entropy encoding and decoding and method therefor |
| CN117917079A (en) * | 2022-07-04 | 2024-04-19 | 嘉兴尚坤科技有限公司 | Method and system for encoding/decoding ultra-high definition video data |
| WO2024148491A1 (en) * | 2023-01-09 | 2024-07-18 | Oppo广东移动通信有限公司 | Encoding method, decoding method, code stream, encoder, decoder and storage medium |
| KR20240114281A (en) * | 2023-01-16 | 2024-07-23 | 주식회사 케이티 | A method of encoding/decoding a video and recording medium storing bitstream |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050012648A1 (en) * | 2003-07-17 | 2005-01-20 | Detlev Marpe | Apparatus and methods for entropy-encoding or entropy-decoding using an initialization of context variables |
| US20050038837A1 (en) * | 2003-07-17 | 2005-02-17 | Detlev Marpe | Method and apparatus for binarization and arithmetic coding of a data value |
| US20090304071A1 (en) * | 2008-06-06 | 2009-12-10 | Apple Inc. | Adaptive application of entropy coding methods |
Family Cites Families (124)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6885864B2 (en) | 1992-03-06 | 2005-04-26 | Aircell, Inc. | Virtual private network for cellular communications |
| US5381145A (en) | 1993-02-10 | 1995-01-10 | Ricoh Corporation | Method and apparatus for parallel decoding and encoding of data |
| US5717394A (en) | 1993-02-10 | 1998-02-10 | Ricoh Company Ltd. | Method and apparatus for encoding and decoding data |
| JPH06334993A (en) * | 1993-03-26 | 1994-12-02 | Sony Corp | Moving picture encoder or decoder and variable length encoding and decoding method for motion vector |
| KR970003102B1 (en) * | 1993-09-17 | 1997-03-14 | 대우전자 주식회사 | Half pixel motion compensation circuit for video decoder |
| JP3560393B2 (en) | 1995-07-06 | 2004-09-02 | 株式会社日鉱マテリアルズ | Manufacturing method of aluminum alloy sputtering target |
| US5847776A (en) | 1996-06-24 | 1998-12-08 | Vdonet Corporation Ltd. | Method for entropy constrained motion estimation and coding of motion vectors with increased search range |
| JP2866616B2 (en) | 1996-06-24 | 1999-03-08 | 昇 大串 | Health bed |
| JPH10151789A (en) | 1996-11-22 | 1998-06-09 | Eiji Shiozaki | Method for controlling powder fly type printing and printing apparatus |
| CN1253652A (en) * | 1997-03-31 | 2000-05-17 | 松下电器产业株式会社 | Dynatic image display method and device therefor |
| KR100255754B1 (en) * | 1997-08-30 | 2000-05-01 | 전주범 | Method for encoding a binary shape signal |
| CN1146245C (en) * | 1997-12-22 | 2004-04-14 | 株式会社大宇电子 | Interlaced binary shape coding method and apparatus |
| US6677868B2 (en) * | 2001-03-16 | 2004-01-13 | Sharp Laboratories Of America, Inc. | Entropy coding with adaptive syntax to replace high probability symbols with lower probabilities symbols |
| US6865298B2 (en) | 2001-03-30 | 2005-03-08 | Sharp Laboratories Of America, Inc. | Compound document compression based upon neighboring pixels |
| JP4125565B2 (en) * | 2001-08-31 | 2008-07-30 | 松下電器産業株式会社 | Image encoding method, image decoding method and apparatus |
| US20030048808A1 (en) * | 2001-09-12 | 2003-03-13 | Stahl Thomas Anthony | Method and apparatus for changing received streaming content channels |
| KR100765060B1 (en) | 2001-11-09 | 2007-10-09 | 주식회사 포스코 | An apparatus for controlling the water drain in sediment |
| US7435543B2 (en) * | 2001-11-27 | 2008-10-14 | Animal Technology Institute Of Taiwan | Genetic markers for pig backfat thickness |
| WO2003053066A1 (en) * | 2001-12-17 | 2003-06-26 | Microsoft Corporation | Skip macroblock coding |
| US7003035B2 (en) * | 2002-01-25 | 2006-02-21 | Microsoft Corporation | Video coding methods and apparatuses |
| JP2003348598A (en) * | 2002-04-12 | 2003-12-05 | Seiko Epson Corp | Method and apparatus for memory efficient compressed domain video processing and for fast inverse motion compensation using factorization and integer approximation |
| JP2003319391A (en) * | 2002-04-26 | 2003-11-07 | Sony Corp | Encoding apparatus and method, decoding apparatus and method, recording medium, and program |
| EP2296282B1 (en) * | 2002-05-02 | 2013-11-06 | Fraunhofer-Gesellschaft zur Förderung der Angewandten Forschung e.V. | method and device for arithmetic encoding and decoding with use of multiple lookup tables |
| PT1467491E (en) | 2002-05-02 | 2007-03-30 | Fraunhofer Ges Forschung | Arithmetical coding of transform coefficients |
| JP4230188B2 (en) * | 2002-06-06 | 2009-02-25 | パナソニック株式会社 | Variable length encoding method and variable length decoding method |
| US6919105B2 (en) | 2003-01-06 | 2005-07-19 | Philip Morris Usa Inc. | Continuous process for retaining solid adsorbent particles on shaped micro-cavity fibers |
| US20050013498A1 (en) * | 2003-07-18 | 2005-01-20 | Microsoft Corporation | Coding of motion vector information |
| US7499495B2 (en) * | 2003-07-18 | 2009-03-03 | Microsoft Corporation | Extended range motion vectors |
| US7233622B2 (en) * | 2003-08-12 | 2007-06-19 | Lsi Corporation | Reduced complexity efficient binarization method and/or circuit for motion vector residuals |
| EP1513350A1 (en) * | 2003-09-03 | 2005-03-09 | Thomson Licensing S.A. | Process and arrangement for encoding video pictures |
| US6771197B1 (en) | 2003-09-26 | 2004-08-03 | Mitsubishi Electric Research Laboratories, Inc. | Quantizing signals using sparse generator factor graph codes |
| US7599435B2 (en) | 2004-01-30 | 2009-10-06 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Video frame encoding and decoding |
| JP4241417B2 (en) * | 2004-02-04 | 2009-03-18 | 日本ビクター株式会社 | Arithmetic decoding device and arithmetic decoding program |
| JP4408766B2 (en) | 2004-07-30 | 2010-02-03 | Ｎｔｎ株式会社 | Wheel bearing device with IC tag |
| CN1589023A (en) * | 2004-08-06 | 2005-03-02 | 联合信源数字音视频技术（北京）有限公司 | Coding and decoding method and device for multiple coded list lengthening based on context |
| US7609947B2 (en) | 2004-09-10 | 2009-10-27 | Panasonic Corporation | Method and apparatus for coordinating playback from multiple video sources |
| CN101052972A (en) * | 2004-09-22 | 2007-10-10 | 液滴技术有限公司 | Multiple technique entropy coding system and method |
| DE102004049156B4 (en) * | 2004-10-08 | 2006-07-13 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Coding scheme for a temporally variable graphic model representing data stream |
| KR20060038189A (en) * | 2004-10-29 | 2006-05-03 | 삼성전자주식회사 | Method and apparatus for context-based adaptive binary arithmetic coding |
| KR100636229B1 (en) * | 2005-01-14 | 2006-10-19 | 학교법인 성균관대학 | Method and apparatus for adaptive entropy encoding and decoding for scalable video coding |
| KR100703760B1 (en) | 2005-03-18 | 2007-04-06 | 삼성전자주식회사 | Video encoding/decoding method using motion prediction between temporal levels and apparatus thereof |
| AU2006201490B2 (en) * | 2005-04-19 | 2008-05-22 | Samsung Electronics Co., Ltd. | Method and apparatus for adaptively selecting context model for entropy coding |
| RU2336661C2 (en) * | 2005-04-19 | 2008-10-20 | Самсунг Электроникс Ко., Лтд. | Method and device of adaptive choice of context model for entropy encoding |
| EP1727371A1 (en) * | 2005-05-27 | 2006-11-29 | Thomson Licensing | Method for controlling the encoder output bit rate in a block-based video encoder, and corresponding video encoder apparatus |
| CN1878309A (en) * | 2005-06-08 | 2006-12-13 | 松下电器产业株式会社 | Video encoding device |
| JP4856954B2 (en) * | 2005-06-08 | 2012-01-18 | パナソニック株式会社 | Image encoding device |
| KR100683788B1 (en) | 2005-06-25 | 2007-02-20 | 삼성에스디아이 주식회사 | Alternative-Current type of discharge display panel wherein electrode lines of plural layers are formed |
| RU2371881C1 (en) * | 2005-07-08 | 2009-10-27 | ЭлДжи ЭЛЕКТРОНИКС ИНК. | Method of modelling video signal coding information for information compression/decompression |
| CN101223780A (en) * | 2005-07-18 | 2008-07-16 | 三星电子株式会社 | Video coding method and apparatus for reducing mismatch between encoder and decoder |
| CN100461863C (en) * | 2005-08-05 | 2009-02-11 | 上海富瀚微电子有限公司 | Context basing self-adaptive binary arithmetic decoder |
| EP1755341A1 (en) * | 2005-08-19 | 2007-02-21 | Texas Instruments Incorporated | Method of quantization of transform coefficients |
| US7221296B2 (en) * | 2005-08-22 | 2007-05-22 | Streaming Networks (Pvt.) Ltd. | Method and system for fast context based adaptive binary arithmetic coding |
| KR100813963B1 (en) | 2005-09-16 | 2008-03-14 | 세종대학교산학협력단 | Method and apparatus for loseless encoding and decoding image |
| CN100403801C (en) * | 2005-09-23 | 2008-07-16 | 联合信源数字音视频技术（北京）有限公司 | Adaptive entropy coding/decoding method based on context |
| CN100466739C (en) | 2005-10-12 | 2009-03-04 | 华为技术有限公司 | CABAC decoding system and method |
| CN101304708B (en) | 2005-10-31 | 2011-06-22 | 生立生物科技股份有限公司 | Sclerotic ring cingulum and manufacturing method thereof |
| KR100873636B1 (en) * | 2005-11-14 | 2008-12-12 | 삼성전자주식회사 | Method and apparatus for encoding/decoding image using single coding mode |
| WO2007059788A1 (en) | 2005-11-25 | 2007-05-31 | Gregor Zebic | Home network application using wirelined and wireless secure links |
| US7245242B2 (en) * | 2005-11-28 | 2007-07-17 | Conexant Systems, Inc. | Decoding systems and methods |
| US7956930B2 (en) | 2006-01-06 | 2011-06-07 | Microsoft Corporation | Resampling and picture resizing operations for multi-resolution video coding and decoding |
| JP4318050B2 (en) | 2006-01-24 | 2009-08-19 | ソニー株式会社 | Wireless channel determination and selection method and access point device |
| KR100750165B1 (en) * | 2006-02-22 | 2007-08-17 | 삼성전자주식회사 | Cabac encoding method and apparatus using improved context model selection with better compression ratio, and cabac decoding method and apparatus |
| US20070217702A1 (en) * | 2006-03-14 | 2007-09-20 | Sung Chih-Ta S | Method and apparatus for decoding digital video stream |
| US8848789B2 (en) | 2006-03-27 | 2014-09-30 | Qualcomm Incorporated | Method and system for coding and decoding information associated with video compression |
| US7929608B2 (en) * | 2006-03-28 | 2011-04-19 | Sony Corporation | Method of reducing computations in intra-prediction and mode decision processes in a digital video encoder |
| JP2007300455A (en) * | 2006-05-01 | 2007-11-15 | Victor Co Of Japan Ltd | Arithmetic encoding apparatus, and context table initialization method in arithmetic encoding apparatus |
| JP2007306370A (en) * | 2006-05-12 | 2007-11-22 | Victor Co Of Japan Ltd | Image coding recorder |
| US8472527B2 (en) * | 2006-09-13 | 2013-06-25 | Texas Instruments Incorporated | Hierarchical motion estimation using original frame for sub-sampled reference |
| CN101198051B (en) * | 2006-12-07 | 2011-10-05 | 深圳艾科创新微电子有限公司 | Method and device for implementing entropy decoder based on H.264 |
| US20080162432A1 (en) * | 2006-12-29 | 2008-07-03 | Wen-Shan Wang | Search table for unary k-th order exp-golomb decoder |
| US7573405B2 (en) * | 2006-12-29 | 2009-08-11 | Texas Instruments Incorporated | System updating accumulated statistics coding variable values after multiple encodings |
| CN101005287A (en) * | 2007-01-26 | 2007-07-25 | 北京中星微电子有限公司 | Method, system and coding device for improving arithmetic coaing speed |
| US20080225947A1 (en) * | 2007-03-13 | 2008-09-18 | Matthias Narroschke | Quantization for hybrid video coding |
| CN101115200B (en) * | 2007-04-20 | 2010-05-19 | 西安交通大学 | Effective motion vector telescopic encoding method |
| WO2008136828A1 (en) * | 2007-05-04 | 2008-11-13 | Qualcomm Incorporated | Video coding mode selection using estimated coding costs |
| JP2008300455A (en) | 2007-05-29 | 2008-12-11 | Sumitomo Electric Ind Ltd | Power module |
| RU2496252C2 (en) * | 2007-06-29 | 2013-10-20 | Шарп Кабусики Кайся | Image coding apparatus, image coding method, image decoding apparatus, image decoding method, program and recording medium |
| US9648325B2 (en) * | 2007-06-30 | 2017-05-09 | Microsoft Technology Licensing, Llc | Video decoding implementations for a graphics processing unit |
| US8254455B2 (en) * | 2007-06-30 | 2012-08-28 | Microsoft Corporation | Computing collocated macroblock information for direct mode macroblocks |
| KR101408698B1 (en) | 2007-07-31 | 2014-06-18 | 삼성전자주식회사 | Method and apparatus for encoding/decoding image using weighted prediction |
| US7839311B2 (en) * | 2007-08-31 | 2010-11-23 | Qualcomm Incorporated | Architecture for multi-stage decoding of a CABAC bitstream |
| US7535387B1 (en) * | 2007-09-10 | 2009-05-19 | Xilinx, Inc. | Methods and systems for implementing context adaptive binary arithmetic coding |
| CN101478672B (en) * | 2008-01-04 | 2012-12-19 | 华为技术有限公司 | Video encoding, decoding method and apparatus, video processing system |
| EP2077551B1 (en) * | 2008-01-04 | 2011-03-02 | Dolby Sweden AB | Audio encoder and decoder |
| JP4915350B2 (en) * | 2008-01-16 | 2012-04-11 | 日本電気株式会社 | Entropy encoder, video encoding device, video encoding method, and video encoding program |
| US8855196B2 (en) | 2008-01-22 | 2014-10-07 | Dolby Laboratories Licensing Corporation | Adaptive motion information cost estimation with dynamic look-up table updating |
| US20090201983A1 (en) | 2008-02-07 | 2009-08-13 | Motorola, Inc. | Method and apparatus for estimating high-band energy in a bandwidth extension system |
| JP4479829B2 (en) | 2008-04-03 | 2010-06-09 | ソニー株式会社 | Imaging apparatus and imaging method |
| US7656320B2 (en) | 2008-04-10 | 2010-02-02 | Red Hat, Inc. | Difference coding adaptive context model using counting |
| JP4962400B2 (en) | 2008-04-30 | 2012-06-27 | ソニー株式会社 | Arithmetic decoding device |
| US8179974B2 (en) * | 2008-05-02 | 2012-05-15 | Microsoft Corporation | Multi-level representation of reordered transform coefficients |
| CN101267559A (en) * | 2008-05-08 | 2008-09-17 | 上海交通大学 | Universal entropy decoding method and device for video decoder |
| CN101304522A (en) * | 2008-06-20 | 2008-11-12 | 中国民航大学 | Considerable information hide method using JPEG2000 compression image as carrier |
| JP2012504925A (en) * | 2008-10-06 | 2012-02-23 | エルジー エレクトロニクス インコーポレイティド | Video signal processing method and apparatus |
| US7932843B2 (en) * | 2008-10-17 | 2011-04-26 | Texas Instruments Incorporated | Parallel CABAC decoding for video decompression |
| WO2010043773A1 (en) * | 2008-10-17 | 2010-04-22 | Nokia Corporation | Sharing of motion vector in 3d video coding |
| EP2182732A1 (en) * | 2008-10-28 | 2010-05-05 | Panasonic Corporation | Switching between scans in image coding |
| US20100127904A1 (en) | 2008-11-26 | 2010-05-27 | Horizon Semiconductors Ltd. | Implementation of a rapid arithmetic binary decoding system of a suffix length |
| AU2009324014A1 (en) * | 2008-12-03 | 2011-06-23 | Nokia Corporation | Switching between DCT coefficient coding modes |
| JP2010268094A (en) * | 2009-05-13 | 2010-11-25 | Sony Corp | Image decoder and image decoding method |
| JP2010278668A (en) | 2009-05-27 | 2010-12-09 | Sony Corp | Coding apparatus and method, and decoding apparatus and method |
| US9479812B2 (en) * | 2009-06-05 | 2016-10-25 | Time Warner Cable Enterprises Llc | User selection of software components in a television set-top box |
| KR101377530B1 (en) * | 2009-08-21 | 2014-03-27 | 에스케이텔레콤 주식회사 | Video Coding Method and Apparatus by Using Adaptive Motion Vector Resolution |
| KR101678968B1 (en) * | 2009-08-21 | 2016-11-25 | 에스케이텔레콤 주식회사 | Reference Picture Interpolation Method and Apparatus and Video Coding Method and Apparatus Using Same |
| CN102055483B (en) * | 2009-10-29 | 2013-05-08 | 鸿富锦精密工业（深圳）有限公司 | Entropy decoding device |
| JP5409268B2 (en) * | 2009-10-30 | 2014-02-05 | 株式会社日立ソリューションズ | Information presenting apparatus and portable terminal |
| KR20110071047A (en) * | 2009-12-20 | 2011-06-28 | 엘지전자 주식회사 | A method and an apparatus for decoding a video signal |
| US9237355B2 (en) | 2010-02-19 | 2016-01-12 | Qualcomm Incorporated | Adaptive motion resolution for video coding |
| US9973768B2 (en) | 2010-03-16 | 2018-05-15 | Texas Instruments Incorporated | CABAC decoder with decoupled arithmetic decoding and inverse binarization |
| CN101841707B (en) * | 2010-03-19 | 2012-01-04 | 西安电子科技大学 | High-speed real-time processing arithmetic coding method based on JPEG 2000 standard |
| CN103119849B (en) * | 2010-04-13 | 2017-06-16 | 弗劳恩霍夫应用研究促进协会 | Probability interval partition encoding device and decoder |
| US20120014429A1 (en) * | 2010-07-15 | 2012-01-19 | Jie Zhao | Methods and Systems for Parallel Video Encoding and Parallel Video Decoding |
| US9456015B2 (en) | 2010-08-10 | 2016-09-27 | Qualcomm Incorporated | Representation groups for network streaming of coded multimedia data |
| CN101938654B (en) * | 2010-08-17 | 2013-04-10 | 浙江大学 | Method and device for optimizing and quantifying conversion coefficients |
| CN101951516B (en) * | 2010-09-25 | 2013-06-05 | 清华大学 | Parallel encoding realization circuit and encoding method based on CABAC (Context-based Adaptive Binary Arithmetic Coding) in H.264/AVC (Advanced Video Coding) |
| US8913666B2 (en) | 2010-10-01 | 2014-12-16 | Qualcomm Incorporated | Entropy coding coefficients using a joint context model |
| US10327008B2 (en) * | 2010-10-13 | 2019-06-18 | Qualcomm Incorporated | Adaptive motion vector resolution signaling for video coding |
| JP5570437B2 (en) * | 2011-01-12 | 2014-08-13 | 三菱電機株式会社 | Entropy encoding device, entropy decoding device, entropy encoding method, and entropy decoding method |
| US9008181B2 (en) | 2011-01-24 | 2015-04-14 | Qualcomm Incorporated | Single reference picture list utilization for interprediction video coding |
| ITTO20110414A1 (en) * | 2011-05-11 | 2012-11-12 | St Microelectronics Pvt Ltd | PROCEDURE AND EQUIPMENT TO PROCESS VIDEO SIGNALS, COMPUTER PRODUCT AND SIGNALING CODIFIED RELATIVE |
| BR112013032333B1 (en) * | 2011-06-16 | 2022-07-26 | Ge Video Compression, Llc | AUXILIARY MODE SWITCHING FOR ENTROPY ENCODING |
| US9654772B2 (en) | 2012-01-19 | 2017-05-16 | Qualcomm Incorporated | Context adaptive entropy coding with a reduced initialization value set |
| US9584802B2 (en) | 2012-04-13 | 2017-02-28 | Texas Instruments Incorporated | Reducing context coded and bypass coded bins to improve context adaptive binary arithmetic coding (CABAC) throughput |
| US10856009B2 (en) * | 2014-09-04 | 2020-12-01 | Mediatek Inc. | Method of block vector clipping and coding for screen content coding and video coding |
-
2012
- 2012-06-18 BR BR112013032333-7A patent/BR112013032333B1/en active IP Right Grant
- 2012-06-18 SI SI201232038T patent/SI3930330T1/en unknown
- 2012-06-18 KR KR1020167003248A patent/KR101730587B1/en active IP Right Grant
- 2012-06-18 CN CN201710694424.6A patent/CN107529708B/en active Active
- 2012-06-18 EP EP20155692.5A patent/EP3703367A1/en active Pending
- 2012-06-18 LT LTEP21189232.8T patent/LT3930330T/en unknown
- 2012-06-18 CN CN201280039832.3A patent/CN103733622B/en active Active
- 2012-06-18 AU AU2012268951A patent/AU2012268951B2/en active Active
- 2012-06-18 EP EP18152613.8A patent/EP3343781B1/en active Active
- 2012-06-18 CN CN201710965903.7A patent/CN107613307B/en active Active
- 2012-06-18 KR KR1020177010779A patent/KR101882848B1/en active IP Right Grant
- 2012-06-18 PL PL18152613T patent/PL3343781T3/en unknown
- 2012-06-18 BR BR112013032332-9A patent/BR112013032332B1/en active IP Right Grant
- 2012-06-18 IL IL312098A patent/IL312098A/en unknown
- 2012-06-18 MX MX2013014867A patent/MX336735B/en unknown
- 2012-06-18 HU HUE21189232A patent/HUE062605T2/en unknown
- 2012-06-18 AU AU2012268950A patent/AU2012268950B2/en active Active
- 2012-06-18 HU HUE18152613A patent/HUE057958T2/en unknown
- 2012-06-18 PL PL12728084.0T patent/PL2721819T3/en unknown
- 2012-06-18 EP EP18203401.7A patent/EP3471415B1/en active Active
- 2012-06-18 RS RS20220220A patent/RS63053B1/en unknown
- 2012-06-18 CN CN201710977462.2A patent/CN107517384B/en active Active
- 2012-06-18 RU RU2017109078A patent/RU2658883C1/en active
- 2012-06-18 KR KR1020167026722A patent/KR101818978B1/en active IP Right Grant
- 2012-06-18 PT PT181526138T patent/PT3343781T/en unknown
- 2012-06-18 RS RS20231075A patent/RS64867B1/en unknown
- 2012-06-18 PL PL18203401T patent/PL3471415T3/en unknown
- 2012-06-18 EP EP23177778.0A patent/EP4277276A1/en active Pending
- 2012-06-18 CA CA2839569A patent/CA2839569C/en active Active
- 2012-06-18 RU RU2014101166A patent/RU2615681C2/en active
- 2012-06-18 BR BR122021000264-9A patent/BR122021000264B1/en active IP Right Grant
- 2012-06-18 KR KR1020187021221A patent/KR102052471B1/en active IP Right Grant
- 2012-06-18 CN CN201710694312.0A patent/CN107801042B/en active Active
- 2012-06-18 KR KR1020147000948A patent/KR101619333B1/en active IP Right Grant
- 2012-06-18 CN CN201710694799.2A patent/CN107529710B/en active Active
- 2012-06-18 ES ES12728084T patent/ES2961663T3/en active Active
- 2012-06-18 ES ES21189232T patent/ES2953667T3/en active Active
- 2012-06-18 ES ES12734832T patent/ES2704627T3/en active Active
- 2012-06-18 CN CN201280039896.3A patent/CN103931194B/en active Active
- 2012-06-18 CN CN201280039922.2A patent/CN103748886B/en active Active
- 2012-06-18 JP JP2014515229A patent/JP6059212B2/en active Active
- 2012-06-18 MY MYPI2014000424A patent/MY170940A/en unknown
- 2012-06-18 ES ES18203401T patent/ES2897209T3/en active Active
- 2012-06-18 BR BR122020024986-2A patent/BR122020024986B1/en active IP Right Grant
- 2012-06-18 IL IL290229A patent/IL290229B2/en unknown
- 2012-06-18 CN CN201710694391.5A patent/CN107347161B/en active Active
- 2012-06-18 CN CN201710477936.7A patent/CN107465928B/en active Active
- 2012-06-18 CN CN201710962467.8A patent/CN107770559B/en active Active
- 2012-06-18 IL IL305221A patent/IL305221B2/en unknown
- 2012-06-18 MX MX2013014868A patent/MX2013014868A/en active IP Right Grant
- 2012-06-18 CN CN201710478537.2A patent/CN107333141B/en active Active
- 2012-06-18 PT PT127280840T patent/PT2721819T/en unknown
- 2012-06-18 UA UAA201600953A patent/UA115186C2/en unknown
- 2012-06-18 IL IL299953A patent/IL299953B2/en unknown
- 2012-06-18 MY MYPI2014000425A patent/MY184473A/en unknown
- 2012-06-18 RS RS20230766A patent/RS64604B1/en unknown
- 2012-06-18 SI SI201232044T patent/SI2721819T1/en unknown
- 2012-06-18 CN CN201710491640.0A patent/CN107333142B/en active Active
- 2012-06-18 CN CN201710477937.1A patent/CN107333140B/en active Active
- 2012-06-18 AP AP2014007361A patent/AP3686A/en active
- 2012-06-18 EP EP12728084.0A patent/EP2721819B1/en active Active
- 2012-06-18 CN CN201710962381.5A patent/CN107517383B/en active Active
- 2012-06-18 IL IL252388A patent/IL252388B/en unknown
- 2012-06-18 CN CN201710694392.XA patent/CN107529705B/en active Active
- 2012-06-18 FI FIEP12728084.0T patent/FI2721819T3/en active
- 2012-06-18 EP EP12734832.4A patent/EP2721822B1/en active Active
- 2012-06-18 IL IL311563A patent/IL311563A/en unknown
- 2012-06-18 EP EP12728524.5A patent/EP2721820A1/en not_active Ceased
- 2012-06-18 KR KR1020147001047A patent/KR101662136B1/en active IP Right Grant
- 2012-06-18 EP EP21189232.8A patent/EP3930330B1/en active Active
- 2012-06-18 KR KR1020207034609A patent/KR102437504B1/en active IP Right Grant
- 2012-06-18 CN CN201710694423.1A patent/CN107529707B/en active Active
- 2012-06-18 KR KR1020197018337A patent/KR102160647B1/en active IP Right Grant
- 2012-06-18 CN CN201710694354.4A patent/CN107529704B/en active Active
- 2012-06-18 RU RU2014101164/08A patent/RU2595934C2/en active
- 2012-06-18 IL IL309379A patent/IL309379B1/en unknown
- 2012-06-18 PL PL12734832T patent/PL2721822T3/en unknown
- 2012-06-18 WO PCT/EP2012/061613 patent/WO2012172113A1/en active Application Filing
- 2012-06-18 FI FIEP21189232.8T patent/FI3930330T3/en active
- 2012-06-18 UA UAA201701791A patent/UA123987C2/en unknown
- 2012-06-18 KR KR1020187000824A patent/KR101994869B1/en active IP Right Grant
- 2012-06-18 IL IL297827A patent/IL297827B2/en unknown
- 2012-06-18 ES ES18152613T patent/ES2906869T3/en active Active
- 2012-06-18 HU HUE18203401A patent/HUE056273T2/en unknown
- 2012-06-18 CA CA2839560A patent/CA2839560C/en active Active
- 2012-06-18 WO PCT/EP2012/061614 patent/WO2012172114A1/en active Application Filing
- 2012-06-18 RS RS20211362A patent/RS62714B1/en unknown
- 2012-06-18 AP AP2014007360A patent/AP4072A/en active
- 2012-06-18 DK DK18203401.7T patent/DK3471415T3/en active
- 2012-06-18 KR KR1020247003097A patent/KR20240017975A/en active Application Filing
- 2012-06-18 JP JP2014515231A patent/JP5952900B2/en active Active
- 2012-06-18 DK DK12728084.0T patent/DK2721819T3/en active
- 2012-06-18 RU RU2016129117A patent/RU2642373C1/en active
- 2012-06-18 WO PCT/EP2012/061615 patent/WO2012172115A1/en active Application Filing
- 2012-06-18 PT PT182034017T patent/PT3471415T/en unknown
- 2012-06-18 SI SI201231980T patent/SI3343781T1/en unknown
- 2012-06-18 LT LTEP18203401.7T patent/LT3471415T/en unknown
- 2012-06-18 CN CN201710695327.9A patent/CN107360423B/en active Active
- 2012-06-18 LT LTEPPCT/EP2012/061615T patent/LT2721819T/en unknown
- 2012-06-18 DK DK12734832.4T patent/DK2721822T3/en active
- 2012-06-18 CN CN201710694268.3A patent/CN107801041B/en active Active
- 2012-06-18 CN CN201710477825.6A patent/CN107465927B/en active Active
- 2012-06-18 DK DK18152613.8T patent/DK3343781T3/en active
- 2012-06-18 KR KR1020207015415A patent/KR102254000B1/en active IP Right Grant
- 2012-06-18 HU HUE12728084A patent/HUE063990T2/en unknown
- 2012-06-18 LT LTEP18152613.8T patent/LT3343781T/en unknown
- 2012-06-18 EP EP21211399.7A patent/EP4033672A1/en active Pending
- 2012-06-18 MX MX2017000909A patent/MX364201B/en unknown
- 2012-06-18 MX MX2016001282A patent/MX345195B/en unknown
- 2012-06-18 SI SI201231961T patent/SI3471415T1/en unknown
- 2012-06-18 CN CN201710694425.0A patent/CN107529709B/en active Active
- 2012-06-18 CN CN201710694930.5A patent/CN107360434B/en active Active
- 2012-06-18 KR KR1020227029290A patent/KR102631388B1/en active IP Right Grant
- 2012-06-18 CN CN201710694292.7A patent/CN107801025B/en active Active
- 2012-06-18 CN CN201710694395.3A patent/CN107529706B/en active Active
- 2012-06-18 PT PT211892328T patent/PT3930330T/en unknown
- 2012-06-18 CN CN201710477823.7A patent/CN107465926B/en active Active
- 2012-06-18 DK DK21189232.8T patent/DK3930330T3/en active
- 2012-06-18 PL PL21189232.8T patent/PL3930330T3/en unknown
- 2012-06-18 PT PT12734832T patent/PT2721822T/en unknown
- 2012-06-18 MY MYPI2018001555A patent/MY202391A/en unknown
- 2012-06-18 IL IL304856A patent/IL304856B2/en unknown
- 2012-06-18 JP JP2014515230A patent/JP5925884B2/en active Active
- 2012-06-18 HR HRP20220272TT patent/HRP20220272T1/en unknown
- 2012-06-18 HU HUE12734832A patent/HUE042350T2/en unknown
- 2012-06-18 KR KR1020197035422A patent/KR102187981B1/en active IP Right Grant
- 2012-06-18 AP AP2016009618A patent/AP2016009618A0/en unknown
-
2013
- 2013-12-16 MX MX2023005264A patent/MX2023005264A/en unknown
- 2013-12-16 CL CL2013003603A patent/CL2013003603A1/en unknown
- 2013-12-16 US US14/108,173 patent/US9918090B2/en active Active
- 2013-12-16 CL CL2013003601A patent/CL2013003601A1/en unknown
- 2013-12-16 MX MX2023005265A patent/MX2023005265A/en unknown
- 2013-12-16 MX MX2019004403A patent/MX2019004403A/en unknown
- 2013-12-16 US US14/108,108 patent/US9473170B2/en active Active
- 2013-12-16 US US14/108,132 patent/US9455744B2/en active Active
- 2013-12-17 IL IL230023A patent/IL230023A/en active IP Right Grant
-
2014
- 2014-01-06 ZA ZA2014/00030A patent/ZA201400030B/en unknown
- 2014-01-06 ZA ZA2014/00029A patent/ZA201400029B/en unknown
- 2014-01-12 IL IL230415A patent/IL230415A/en not_active IP Right Cessation
- 2014-01-16 CO CO14007521A patent/CO6852030A2/en unknown
- 2014-01-16 CO CO14007528A patent/CO6852031A2/en active IP Right Grant
- 2014-10-23 HK HK14110575.6A patent/HK1197128A1/en unknown
-
2016
- 2016-04-20 JP JP2016084671A patent/JP6356723B2/en active Active
- 2016-04-26 AU AU2016202638A patent/AU2016202638B2/en active Active
- 2016-08-16 US US15/238,523 patent/US9729883B2/en active Active
- 2016-08-16 US US15/238,294 patent/US9596475B2/en active Active
- 2016-08-23 US US15/244,126 patent/US9628827B2/en active Active
- 2016-08-23 US US15/244,156 patent/US9686568B2/en active Active
- 2016-12-08 JP JP2016238858A patent/JP6492047B2/en active Active
- 2016-12-19 IL IL249644A patent/IL249644A/en active IP Right Grant
-
2017
- 2017-01-30 US US15/419,006 patent/US9743090B2/en active Active
- 2017-02-13 US US15/430,826 patent/US9762913B2/en active Active
- 2017-05-15 US US15/595,097 patent/US9768804B1/en active Active
- 2017-06-28 US US15/636,476 patent/US9930370B2/en active Active
- 2017-06-28 US US15/636,451 patent/US10021393B2/en active Active
- 2017-06-28 US US15/636,425 patent/US9918104B2/en active Active
- 2017-07-05 US US15/641,992 patent/US9936227B2/en active Active
- 2017-07-05 US US15/642,085 patent/US9930371B2/en active Active
- 2017-08-10 US US15/674,150 patent/US9973761B2/en active Active
- 2017-09-03 IL IL254274A patent/IL254274B/en active IP Right Grant
- 2017-09-14 AU AU2017228613A patent/AU2017228613B2/en active Active
- 2017-12-15 US US15/843,679 patent/US10057603B2/en active Active
-
2018
- 2018-01-09 RU RU2018100106A patent/RU2699677C2/en active
- 2018-01-26 US US15/880,837 patent/US10063858B2/en active Active
- 2018-01-26 US US15/880,772 patent/US10148962B2/en active Active
- 2018-01-30 AU AU2018200699A patent/AU2018200699B2/en active Active
- 2018-02-14 PH PH12018500331A patent/PH12018500331A1/en unknown
- 2018-02-14 PH PH12018500330A patent/PH12018500330B1/en unknown
- 2018-02-14 PH PH12018500329A patent/PH12018500329B1/en unknown
- 2018-02-14 PH PH12018500328A patent/PH12018500328B1/en unknown
- 2018-02-15 PH PH12018500355A patent/PH12018500355B1/en unknown
- 2018-02-15 PH PH12018500348A patent/PH12018500348B1/en unknown
- 2018-02-15 PH PH12018500353A patent/PH12018500353B1/en unknown
- 2018-02-15 PH PH12018500349A patent/PH12018500349A1/en unknown
- 2018-02-15 PH PH12018500351A patent/PH12018500351B1/en unknown
- 2018-02-15 PH PH12018500352A patent/PH12018500352B1/en unknown
- 2018-02-15 PH PH12018500350A patent/PH12018500350B1/en unknown
- 2018-02-15 PH PH12018500354A patent/PH12018500354A1/en unknown
- 2018-02-15 PH PH12018500356A patent/PH12018500356B1/en unknown
- 2018-03-12 US US15/918,105 patent/US10440364B2/en active Active
- 2018-04-02 JP JP2018070521A patent/JP6560393B2/en active Active
- 2018-05-02 HK HK18105679.7A patent/HK1246541A1/en unknown
- 2018-05-02 HK HK18105680.4A patent/HK1246542A1/en unknown
- 2018-05-02 HK HK18105678.8A patent/HK1246540A1/en unknown
- 2018-05-03 HK HK18105710.8A patent/HK1246543A1/en unknown
- 2018-05-03 HK HK18105749.3A patent/HK1246545A1/en unknown
- 2018-05-24 HK HK18106776.7A patent/HK1247488A1/en unknown
- 2018-05-24 HK HK18106770.3A patent/HK1247487A1/en unknown
- 2018-05-24 HK HK18106785.6A patent/HK1247489A1/en unknown
- 2018-05-25 HK HK18106812.3A patent/HK1247490A1/en unknown
- 2018-06-03 IL IL259768A patent/IL259768A/en active IP Right Grant
- 2018-06-06 RU RU2018120845A patent/RU2758981C2/en active
- 2018-06-10 HK HK18107525.9A patent/HK1248434B/en unknown
- 2018-06-12 US US16/006,288 patent/US10298964B2/en active Active
- 2018-06-14 HK HK18107723.9A patent/HK1248436A1/en unknown
- 2018-06-27 JP JP2018121555A patent/JP6867334B2/en active Active
- 2018-06-28 HK HK18108377.6A patent/HK1248947A1/en unknown
- 2018-06-28 HK HK18108379.4A patent/HK1248949A1/en unknown
- 2018-06-28 HK HK18108378.5A patent/HK1248948B/en unknown
- 2018-07-03 HK HK18108563.0A patent/HK1249313B/en unknown
- 2018-07-04 HK HK18108644.3A patent/HK1248951A1/en unknown
- 2018-07-17 US US16/037,914 patent/US10313672B2/en active Active
- 2018-07-20 US US16/040,764 patent/US10230954B2/en active Active
- 2018-08-14 US US16/103,266 patent/US10306232B2/en active Active
- 2018-11-19 JP JP2018216486A patent/JP6814780B2/en active Active
-
2019
- 2019-01-02 US US16/238,013 patent/US10425644B2/en active Active
- 2019-01-18 AU AU2019200328A patent/AU2019200328B2/en active Active
- 2019-01-28 US US16/259,738 patent/US10432939B2/en active Active
- 2019-01-28 US US16/259,815 patent/US10432940B2/en active Active
- 2019-03-26 IL IL265636A patent/IL265636B/en active IP Right Grant
- 2019-04-11 PH PH12019500792A patent/PH12019500792A1/en unknown
- 2019-04-11 PH PH12019500790A patent/PH12019500790A1/en unknown
- 2019-05-30 AU AU2019203792A patent/AU2019203792B2/en active Active
- 2019-06-27 US US16/454,247 patent/US10630987B2/en active Active
- 2019-06-27 US US16/454,387 patent/US10630988B2/en active Active
- 2019-08-28 RU RU2019127048A patent/RU2755020C2/en active
- 2019-11-25 US US16/693,886 patent/US10819982B2/en active Active
-
2020
- 2020-01-31 US US16/778,048 patent/US11533485B2/en active Active
- 2020-05-05 US US16/867,149 patent/US11277614B2/en active Active
- 2020-10-23 JP JP2020178469A patent/JP7231594B2/en active Active
-
2021
- 2021-04-08 JP JP2021065563A patent/JP7100836B2/en active Active
- 2021-06-02 IL IL283649A patent/IL283649B/en unknown
- 2021-07-29 AU AU2021209284A patent/AU2021209284B2/en active Active
- 2021-11-03 CY CY20211100947T patent/CY1124687T1/en unknown
-
2022
- 2022-02-01 US US17/590,596 patent/US11838511B2/en active Active
- 2022-06-30 IL IL294454A patent/IL294454B2/en unknown
- 2022-09-09 US US17/930,997 patent/US12069267B2/en active Active
- 2022-09-19 AU AU2022235513A patent/AU2022235513B2/en active Active
- 2022-11-07 US US18/053,219 patent/US20230188719A1/en active Pending
-
2023
- 2023-02-16 JP JP2023022627A patent/JP7464768B2/en active Active
- 2023-06-16 AU AU2023203785A patent/AU2023203785B2/en active Active
- 2023-10-06 US US18/482,683 patent/US20240048708A1/en active Pending
-
2024
- 2024-03-15 AU AU2024201731A patent/AU2024201731B2/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050012648A1 (en) * | 2003-07-17 | 2005-01-20 | Detlev Marpe | Apparatus and methods for entropy-encoding or entropy-decoding using an initialization of context variables |
| US20050038837A1 (en) * | 2003-07-17 | 2005-02-17 | Detlev Marpe | Method and apparatus for binarization and arithmetic coding of a data value |
| US20090304071A1 (en) * | 2008-06-06 | 2009-12-10 | Apple Inc. | Adaptive application of entropy coding methods |
Non-Patent Citations (1)
| Title |
|---|
| VIVIENNE SZE: "Simplified MVD context selection (Extension of JCTVC-E324)", 《JOINT COLLABORATIVE TEAM ON VIDEO CODING (JCT-VC) OF ITU-T SG16 WP3 AND ISO/IEC JTC1/SC29/WG11》 * |
Also Published As
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103931194B (en) | The entropy code of motion vector difference |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1248948 Country of ref document: HK |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |